DESCRIPTION TITLE METHOD AND APPARATUS FOR INCREASING A PERFORMANCE OF A PHOTO- VOLTAIC SYSTEM TECHNICAL FIELD The invention relates to a computer-implemented method and apparatus for providing assistance in increasing a perfor- mance of a photovoltaic system and in locating errors within a photovoltaic system. BACKGROUND A photovoltaic system can comprise a plurality of photovol- taic modules which are adapted to convert solar radiation into electrical DC power. The generated DC power is supplied by the photovoltaic panel to an associated DC/AC converter, a so-called inverter, adapted to convert the DC power into electrical AC power. The AC output of the photovoltaic in- verter can be applied to a power grid or used for power dis- tribution to a customer structure, i.e. a residential, com- mercial or industrial or local power supply network. The in- verter can comprise a controller configured to monitor and control photovoltaic modules and to communicate with other components of the photovoltaic system. A photovoltaic system can comprise several inverters each connected to an associ- ated photovoltaic array. The photovoltaic array can comprise several strings each including serial connected photovoltaic modules. Each photovoltaic module comprises photovoltaic cells. A photovoltaic system can comprise several components which build up the system according to a system’s configura- tion. The photovoltaic system can comprise different types of com- ponents such as different photovoltaic modules, different types of DC/DC converters and/or DC/AC converters. Depending on the configuration of the photovoltaic system and depending on environmental influences, the performance of a photovol- taic system can vary significantly. The performance of a pho- tovoltaic system reflects the portion of the solar power re- ceived by the photovoltaic modules being transformed by the photovoltaic system into AC power.

A photovoltaic system can also be connected to a power supply grid to exchange electrical power. Over time, the performance of an installed photovoltaic system can degrade. Typically, the degradation of the performance of a photovoltaic system is caused by aging of the electrical components of the photo- voltaic system, dust forming on the surfaces of the photovol- taic panels or shading of the solar radiation, for instance by trees. For example, partial shading of photovoltaic panels can be caused by dust, snow, vegetation or buildings. Fur- ther, there can be degradations by component failures such as aging DC/DC converters or aging components of AC/DC power in- verters. Also installation faults can lead to a degradation of a photovoltaic system such as an erroneous configuration of the photovoltaic panels or wiring. Moreover, seasonal in- fluences or climate changes or special whether conditions such as hail storms can have an impact on the performance of a photovoltaic system.

The momentary weather conditions have a high impact on the performance of a photovoltaic system. Typically, on cloudy days, the performance of a photovoltaic system is signifi- cantly reduced. Another factor influencing the performance of a photovoltaic system is the operation time of the photovol- taic system. Usually at night, the power generated by a pho- tovoltaic system is almost reduced to zero, whereas under broad daylight the power generated by a photovoltaic system is quite high. Accordingly, there is a variety of different factors which can influence the performance of a photovoltaic system. An operator of a photovoltaic system strives to achieve a high performance of the operated photovoltaic sys- tem for a given infrastructure. In many scenarios, undetected performance degrading factors will lead to a loss of gener- ated power.

US 11 146212 B1 discloses a method and system for solar panel performance monitoring. Further, a system and method for fault identification and alerting are disclosed in US 2020/0127604 A1.

SUMMARY OF THE INVENTION

It is one of the objects of the present invention to provide a computer-implemented method for increasing a performance of a photovoltaic system by error and failure detection.

This object is achieved according to a first aspect of the present invention by a computer-implemented method for providing assistance in increasing a performance of a photo- voltaic system.

According to the first aspect of the invention a computer-im- plemented method for providing assistance in increasing a performance of a photovoltaic system comprises: processing performance relevant data received from heteroge- neous data sources to monitor a performance of the photovol- taic system, comparing the monitored performance with a ref- erence performance, e.g. reference performance data, to de- tect a degradation of said photovoltaic system, processing the received performance relevant data to determine automati- cally a degradation root cause of a detected degradation of the photovoltaic system, determining at least one recommenda- tion to overcome the determined root cause of the detected degradation of the photovoltaic system, and outputting the determined recommendation via a user interface to an operator of the photovoltaic system and/or via a control interface to a controller of said photovoltaic system. The degradation root cause is determined based on a system data model of the monitored photovoltaic system derived from system configura- tion data of the monitored photovoltaic system, and derived from component related data of system components of said pho- tovoltaic system, the configuration data of the monitored photovoltaic system indicating assembly and interconnection between the different system components of the respective photovoltaic system, wherein the component related data in- cludes datasets for each system component of the photovoltaic system, and wherein the system data model of the investigated monitored photovoltaic system is created based on the da- tasets of the system components and by using the configura- tion data of the respective photovoltaic system

With the computer-implemented method according to the present invention, a benchmarking of a monitored performance of a photovoltaic system with a reference performance is per- formed. In this way, a degradation of performance of the mon- itored photovoltaic system can be detected at an early stage.

Moreover, in a preferred embodiment of the computer-imple- mented method according to the present invention does not only detect a degradation of the photovoltaic system but is also capable of determining a root cause for the detected degradation based on the performance relevant data.

Furthermore, the computer-implemented method according to the present invention determines at least one recommendation to overcome a determined root cause output to an operator of the photovoltaic system or output to a controller of the photo- voltaic system. Based on the recommendation, the performance of the photovoltaic system can be increased. The computer-im- plemented method according to the present invention makes use of a plurality of heterogeneous data sources, i.e. various different data sources that deliver different types of infor- mation, to provide a holistic view on the monitored photovol- taic system.

The heterogeneous data sources can comprise internal data sources located at the monitored photovoltaic system and/or external data sources, e.g. data sources of service provid- ers.

The performance relevant data provided by internal data sources located at the monitored photovoltaic system may com- prise, in a possible embodiment, system configuration data indicating a configuration or assembly of system components of the monitored photovoltaic system and their connections and interfaces.

The system configuration data may, for instance, include or indicate at least one of the following: a type of the used photovoltaic panels and their current-voltage curve charac- teristica, type and characteristic of implemented DC/DC power converters, type characteristica of employed DC/AC convert- ers, type and characteristic of implemented MPP trackers, a wiring implemented between the different components of the photovoltaic system according to a wiring plan or scheme. These data may also be stored in an external data source or in a system component library of a local data base as compo- nent related data. The configuration data may further in- clude, for example, an orientation, i.e. with respect to North, and tilt angle of the photovoltaic panels.

The performance relevant data provided by the internal data sources located at the monitored photovoltaic system can also include, according to some embodiments of the computer-imple- mented method, service messages and state codes generated by components of the photovoltaic systems over time, in particu- lar, service messages generated by controllers integrated in components of the photovoltaic system. In a further possible embodiment of the computer-implemented method, the performance relevant data provided by the inter- nal data sources located at the monitored photovoltaic system can also comprise flow data indicating a power generation of the monitored photovoltaic system.

In a still further possible embodiment of the computer-imple- mented method according to the first aspect of the present invention, the performance relevant data provided by data sources located at the monitored photovoltaic system can also comprise irradiation data indicating the solar radiation power at the location of the monitored photovoltaic system.

In a still further possible embodiment of the computer-imple- mented method, the performance relevant data provided by the internal data sources located at the monitored photovoltaic system can also comprise historic performance data related to a performance parameter of the monitored photovoltaic system observed during past observation time periods.

Besides the internal data sources, the computer-implemented method can also process performance relevant data from other heterogeneous data sources, in particular external data sources of service providers.

In a possible embodiment of the computer-implemented method, the performance relevant data provided by external data sources comprise environmental data indicating environmental conditions at the location of the monitored photovoltaic sys- tem over time. These environmental data can comprise weather data concerning the weather at the location of the monitored photovoltaic system, irradiation data indicating the solar power irradiating on the photovoltaic panels of the photovol- taic system, shading data and/or location data concerning the location of the monitored photovoltaic system. In a further possible embodiment of the computer-implemented method, the performance relevant data provided by the exter- nal data sources can also comprise statistical data. This statistical data can include failure and/or error probabili- ties of system components implemented in the monitored photo- voltaic system according to its configuration and/or climate zone data related to a climate zone where the monitored pho- tovoltaic system is located.

In a possible embodiment of the computer-implemented method, the reference performance used to detect a degradation of the monitored photovoltaic system comprises a performance of a reference photovoltaic system.

In a possible embodiment of the computer-implemented method, the reference photovoltaic system is selected from a prede- fined group of photovoltaic systems.

The selection can, for instance, be performed by an operator of the monitored photovoltaic system of interest. A prede- fined group of selectable photovoltaic systems can include neighboring photovoltaic systems located within a predefined range of the geographic location of the monitored photovol- taic system. A selection by the operator may be made, for ex- ample, via an user interface of the photovoltaic system.

A predefined group of selectable photovoltaic systems may, for example, comprise a photovoltaic system having a similar or identical configuration as the monitored photovoltaic sys- tem.

Further, the predefined group of selectable photovoltaic sys- tems can also comprise photovoltaic systems being exposed to similar environmental conditions as the monitored photovol- taic system. For example, photovoltaic systems being exposed to similar environmental conditions as the monitored photo- voltaic system may be photovoltaic systems that are installed within predefined distance range of the monitored photovol- taic system, e.g. systems within a radius in a range between 5 km to 30 km around the monitored photovoltaic system.

In a further possible embodiment of the computer-implemented method, the reference performance used for comparison with the monitored performance of the investigated monitored pho- tovoltaic system to detect a degradation of the monitored photovoltaic system can also comprise an average performance calculated from performances provided by a group of selected photovoltaic systems meeting specified criteria such as being in the vicinity to the geographic location of the monitored photovoltaic system, having a high configuration similarity with the configuration of the monitored photovoltaic system and/or a similarity with the environmental conditions to which the photovoltaic systems are exposed.

In a further possible embodiment of the computer-implemented method, the reference performance used for comparison with the monitored performance of the monitored photovoltaic sys- tem is provided by a simulator based on performance relevant data provided by the heterogeneous data sources, and based on a data model of the monitored photovoltaic system which can be derived from the configuration data of the monitored pho- tovoltaic system. The simulator may be defined as a digital twin (during monitoring) or as a setpoint for performance af- ter installation. For example, the digital model may receive as input the configuration data of the photovoltaic system, i.e. data indicating the electric components of the photovol- taic system and data characterizing performance of the elec- tric components dependent on solar irradiation, a wiring scheme defining the electrical connection between the elec- tric components of the photovoltaic system, and actual weather data, i.e. data on solar irradiation, of the geo- graphic location where the photovoltaic system is installed. The digital model may be configured to determine a reference performance of the photovoltaic system based on the input and deliver, as output, the reference performance.

In a still further possible embodiment of the computer-imple- mented method, the reference performance can also comprise a historic performance of the monitored photovoltaic system in previous observation time periods derived from historic per- formance data related to the power generation by the moni- tored photovoltaic system within the previous observation time periods.

Accordingly, the momentary performance of the monitored pho- tovoltaic system is compared, in this case, with the historic performance of the same photovoltaic system in previous ob- servation time periods. This allows to observe trends in the performance of a monitored photovoltaic system over time.

In a still further possible embodiment of the computer-imple- mented method, the performance of the monitored photovoltaic system is compared with one or more reference performances in parallel or simultaneously to detect a degradation of the monitored photovoltaic system at an early stage of the degra- dation.

In a possible embodiment of the computer-implemented method, different kinds of degradations of a monitored photovoltaic system can be observed.

In a possible embodiment of the computer-implemented method according to the first aspect of the present invention, the detected degradation of a monitored photovoltaic system com- prises a voltage degradation and/or power degradation of the monitored photovoltaic system. For example, the actual deliv- ered voltage and/or power may be lower than expected based on the reference performance. In a further possible embodiment of the computer-implemented method, the detected degradation of the monitored photovol- taic system comprises a degradation with relation to one or more performance factors of the monitored photovoltaic sys- tem.

In a still further possible embodiment of the computer-imple- mented method, the internal data sources of the photovoltaic system are associated to different inverters of the monitored photovoltaic system to provide inverter specific performance relevant data of the monitored photovoltaic system. For exam- ple, each inverter of the photovoltaic system may include a data logger or a similar memory configured to at least tempo- rarily store performance relevant data of the respective in- verter. For example, inverter specific performance data may be state codes representing a functional state of the in- verter, a DC voltage applied to a DC side of the inverter and an AC voltage output by the inverter, a DC power received and an AC power output by the inverter, or similar. By resolving performance data to an inverter level in the way as described above, more accurate data for determining degradation is available and a degradation root cause can be determined even more precisely and reliable.

In a still further possible embodiment of the computer-imple- mented method, a connection status of the different heteroge- neous data sources is continuously monitored to assess a re- liability of the determined recommendations.

In a further possible embodiment of the computer-implemented method, a degradation root cause is determined based on sys- tem data of the monitored photovoltaic system.

System data of the monitored photovoltaic system can be de- rived from system configuration data of the monitored photo- voltaic system received from an external data source or loaded from a database and derived from component related data of system components of the photovoltaic system loaded from a system component library.

In a still further possible embodiment of the computer-imple- mented method, a mapping of the determined degradation root cause to one or more recommendations is performed based on data entries in a mapping table or look-up table stored in a database of a performance center server.

According to a second aspect of the invention a performance center cloud server comprises a performance evaluation engine is adapted to process data received from heterogeneous data sources to monitor a performance of a photovoltaic system, and is adapted to compare the monitored performance with a reference performance to detect a degradation of the respec- tive monitored photovoltaic system. The performance center cloud server further comprises a recommendation engine adapted to process the data received from the heterogeneous data sources and comparison results of the performance evalu- ation engine to determine a degradation root cause of the degradation of the respective photovoltaic system, to deter- mine at least one recommendation to overcome the determined degradation root cause of the detected degradation of the re- spective monitored photovoltaic system, and to output the de- termined recommendation via a user interface to an operator of the photovoltaic system and/or via a control interface to a controller of the respective photovoltaic system. Further, the recommendation engine is adapted to determine the degra- dation root cause based on a system data model of the moni- tored photovoltaic system derived from system configuration data of the monitored photovoltaic system and derived from component related data of system components of said photovol- taic system, optionally loaded from a system component li- brary, the configuration data of the monitored photovoltaic system indicating assembly and interconnection between the different system components of the respective photovoltaic system, wherein the component related data includes datasets for each system component of the photovoltaic system, and wherein the system data model of the investigated monitored photovoltaic system is created based on the datasets of the system components and by using the configuration data of the respective photovoltaic system.

In particular, the performance center cloud server comprises may be configured to perform the method of the first aspect of the invention.

Optionally, the system data model can also use data from ex- ternal data sources, in addition to the system configuration data and component related data. For example, environmental data such as weather data, irradiation data, data regarding shading, e.g. satellite data, visual recordings, can be pro- vided as data from external data sources.

The features and advantages disclosed in connection with one aspect of the invention are also disclosed for the other as- pect and vice versa.

BRIEF DESCRIPTION OF THE DRAWINGS

In the following, possible embodiments of the different as- pects of the present invention are described in more detail with reference to the enclosed figures.

Fig. 1 schematically shows a cloud-based performance cen- ter adapted to perform a computer-implemented method according to an embodiment of the present invention;

Fig. 2 shows a flowchart for illustrating the main steps performed by a computer-implemented method accord- ing to an embodiment of the present invention; Fig. 3 shows schematically a cloud-based system for per- forming the computer-implemented method according an embodiment of the present invention;

Fig. 4 shows an exemplary implementation of a photovoltaic system.

Fig. 1 shows schematically a cloud-based system adapted to perform a computer-implemented method for increasing a per- formance of a photovoltaic system. The cloud-based system comprises a performance center cloud server 1, as illustrated schematically in Fig. 1, comprises a performance evaluation engine 1A and a recommendation engine 1B. The performance center cloud server 1 is connected via a cloud 2 forming a distributed data network to several photovoltaic systems 3-1, 3-2,.,.3-N as shown in Fig. 1. Each photovoltaic system 3-1, 3- 2,...3-N can comprise its internal data sources 4-1, 4-2,... 4-N as shown schematically in Fig. 1. Besides the internal data sources 4-1, 4-2,... 4-N, the cloud 2 may comprise or provide access to external heterogeneous data sources 5-1, 5-2,.,.5-M as shown in Fig. 1. The performance center cloud server 1 can also have access to local databases 6 as illustrated in Fig. 1.

The performance center cloud server 1 can be adapted to exe- cute a computer-implemented method used to search root causes for underperformance. By addressing the found root causes a performance of a monitored photovoltaic system 3 can be in- creased. As mentioned above, the performance center cloud server 1 of the performance center can comprise a performance evaluation engine 1A and a recommendation engine 1B, as shown schematically in Fig. 1.

The performance evaluation engine 1A is adapted to process data received from heterogeneous data sources to monitor a performance of a specific photovoltaic system 3 on investiga- tion and is further adapted to compare the monitored perfor- mance of the investigated photovoltaic system 3 with a refer ence performance to detect a degradation of the monitored in vestigated photovoltaic system 3.

A photovoltaic system 3 can comprise a setup as illustrated in the example of Fig.4. The setup may comprise configurable and non-configurable components. A photovoltaic system 3 can comprise several module fields with photovoltaic modules M connected in series to form strings connected to an MPP tracker of an inverter. "MPP" is an abbrevation for the term

"Maximum Power Point". The number and types of inverters and of the module fields and of other components as well as the wiring between these and other electric components of the photovoltaic system 3 can vary depending on the use case. The individual configuration of the photovoltaic system 3 can be stored in a configuration database of the photovoltaic system 3. Also configuration changes performed at a photovoltaic system 3 stepwise over time can be stored as historic config- uration data reflecting the performed changes. In this way an impact of a configuration change can be found as a possible root cause for a performance degradation or performance en- hancement of the respective photovoltaic system 3.

The recommendation engine 1B of the performance center cloud server 1 is adapted to process data received from the hetero- geneous data sources and comparison results received from the performance evaluation engine 1A to determine a degradation root cause DRC of a degradation of the respective monitored photovoltaic system 3 and is adapted to determine at least one recommendation REC to overcome the determined degradation root cause DRC of the detected degradation of the monitored photovoltaic system 3. The recommendation engine 1B is fur- ther adapted to output the determined recommendation REC via a user interface UI of a user equipment 7 to an operator or user U of the monitored photovoltaic system 3 as shown sche- matically in Fig. 1 and/or via a control interface to a con- troller of the respective photovoltaic system 3.

Fig. 2 shows a flowchart for illustrating the main steps per- formed by a computer-implemented method. As exemplarily il- lustrated in the flowchart of Fig. 2, the computer-imple- mented method comprises five main steps S1 to S5. The com- puter-implemented method as illustrated in Fig. 2 can be per- formed by a program executed by one or more processors of the performance center cloud server 1 as illustrated in the block diagram of Fig. 1.

In a first step S1, performance relevant data received from heterogeneous data sources can be processed to monitor a per- formance of an investigated monitored photovoltaic system 3. The investigated monitored photovoltaic system 3 may be one of the photovoltaic systems 3-1, 3-2 to 3-N as illustrated in Fig. 1. For instance, an operator or user of several photo- voltaic systems 3 may choose to monitor the performance of the second photovoltaic system 3-2 as illustrated in Fig. 1. Selection of the photovoltaic system to be investigated among photovoltaic systems 3-1, 3-2 to 3-N may be made by the oper- ator or user via the user interface UI.

In a further step S2, the monitored performance of the inves- tigated photovoltaic system 3 is compared with a reference performance to detect a degradation of the monitored photo- voltaic system 3 under investigation. The used reference per- formance can comprise a performance of a reference photovol- taic system 3 or a performance of multiple reference photo- voltaic systems 3. Reference can also be a setpoint calcu- lated based on irradiation data (Performance ratio calcula- tion). A reference photovoltaic system 3 can, for example, be selected from a predefined group of photovoltaic systems 3 according to selection criteria. In an example, the prede- fined group of photovoltaic systems 3 can include neighboring photovoltaic systems 3 located within a predefined range of the geographic location of the investigated photovoltaic sys- tem 3, for example, within a range between 5 km and 30 km. Further, the predefined group of photovoltaic systems 3 can also comprise photovoltaic systems 3 having a similar or identical configuration as the investigated monitored photo- voltaic system 3. Further, the predefined group of photovol- taic systems 3 can also comprise photovoltaic systems 3 being exposed to similar environmental conditions as the monitored photovoltaic system 3. Further, the predefined group of pho- tovoltaic systems 3 can also comprise photovoltaic systems 3 having been installed by the same installer as the monitored photovoltaic system 3. For instance, a reference photovoltaic system 3 can comprise a photovoltaic system being close to the geographic position or coordinates of the investigated photovoltaic system 3 and/or having an identical configura- tion of system components as the investigated photovoltaic system 3 and being also exposed to similar environmental con- ditions as the investigated photovoltaic system 3.

Accordingly, in step S2, a benchmarking with a reference per- formance is done to detect a possible degradation of the in- vestigated photovoltaic system 3. As a reference performance, for example, a performance of a reference photovoltaic system 3 can be used. Alternatively, the reference performance may be provided or calculated by a simulator executed by the per- formance evaluation engine 1A of the performance center cloud server 1. The computation can be based on the performance relevant data provided by heterogeneous data sources and on the basis of a configuration data of the monitored photovol- taic system 3 derived from the system configuration data of the respective monitored photovoltaic system 3. An impact of intended configuration changes can be calculated by the simu- lator or by a trained artificial intelligence network of the performance center cloud server 1 to provide recommendations with respect to configuration changes to be performed at the investigated photovoltaic system 3. It may also be provided that the reference performance used in the degradation detection in step S2 includes a historic performance of the same monitored photovoltaic system 3 in previous observation time periods derived from historic per- formance data related to the power generation by the moni- tored photovoltaic system within the previous observation time periods. With the computer-implemented method as illus- trated in the flowchart of Fig. 2, one or more reference per- formances can be used for the comparison of the monitored performance of the investigated photovoltaic system 3 to de- tect a possible degradation of the monitored photovoltaic system 3. Also currents and voltages can be used for compari- son. Accordingly, a benchmarking with one or more reference performances can be performed simultaneously, i.e. in paral- lel.

After having detected a degradation of the photovoltaic sys- tem 3 in step S2, a root cause detection is performed in step S3. In step S3, the performance relevant data received from the heterogeneous data sources can be processed to determine a degradation root cause, abbreviated as DRC in the follow- ing, of the detected degradation of the investigated photo- voltaic system 3. In step S3, a root cause analysis algorithm is performed to detect the original cause for the observed degradation. The observed degradation can comprise a degrada- tion with relation to a predefined key performance indicator or performance factor.

For example, the DRC can be determined based on a system data model of the monitored photovoltaic system 3 under investiga- tion. This system model can be derived in a possible imple- mentation from system configuration data of the monitored photovoltaic system 3 received from an external data source or loaded from a database 6 of the cloud system or received one of the internal data sources 4-1, 4-2, 4-3. The system data model can, for example, be derived from component re- lated data of system components of the photovoltaic system 3. The component related data may, for example, be loaded from a system component library stored in a database 6. The system components of the investigated photovoltaic system can com- prise different analog or digital hardware components and also implemented software modules. The hardware components can for instance comprise photovoltaic panels or modules, the implemented DC/DC inverters, the DC/AC inverters of the pho- tovoltaic system 3 as well as components used for wiring other components of the photovoltaic system 3. For each sys- tem component of the photovoltaic system 3, corresponding da- tasets can be stored in a system component library. Based on the datasets of the system components, a system data model of the investigated monitored photovoltaic system 3 can be cre- ated. The configuration data of the respective photovoltaic system 3 indicates the assembly and interconnection between the different system components of the respective photovol- taic system 3. The data indicating assembly of the system components may include, for example, a tilt angle of the so- lar panels and an orientation of the solar panels with re- spect to North. The data on interconnection contained in the configuration data may include, for example, a wiring scheme of the system components. The generated system data model of the photovoltaic system 3 can be stored in the database 6 to perform a degradation root cause analysis for the respective photovoltaic system 3. The result of the root cause detection analysis is the determination of a specific DRC for the deg- radation of the photovoltaic system 3 detected in step S2.

In a further step S4, at least one recommendation REC to overcome the DRC of the detected degradation of the photovol- taic system 3 is determined. The at least one recommendation can be determined by the recommendation engine 1B of the per- formance center cloud server 1 illustrated in Fig. 1. For ex- ample, a mapping of the determined degradation root cause (DRC) to one or more recommendations REC can be performed in step S4 based on the data entries of a mapping table (MT) stored in the database 6 of the performance center cloud server 1. Optionally, for each possible DRC one or more rec- ommendations REC are stored in a look-up table MT and can be processed to retrieve recommendations for a determined degra- dation root cause. The calculated recommendation REC can also be enriched based on performance relevant data received from the heterogeneous data sources and based on comparison re- sults. The recommendation service provided by the recommenda- tion engine comprises an algorithm that derives a call to ac- tion based on calculated key performance indicators as well as based on data from external data sources (AIs) including recommendation tables.

The system may be triggered, for example, by different peri- odically performed calculations: a) Inverter level energy comparison: Trend analysis of the daily specific yield values of each inverter in the PV system 3. Interpretation directly as alert for unusual deviations between inverters as well as a trigger for further analysis (degradation calculation). b) Neighborhood comparison: Classification of PV systems 3 based on their geographic location and configuration (at least orientation, tilt angle). Analysis of historic perfor- mance data of the previously classified PV systems to detect PV systems 3 that are under-performing compared to other, comparable PV systems 3 in the geographic proximity. Inter- pretation directly as an alert for under-performance of the PV system 3 as well as a trigger for further analysis. c) Performance ratio: Calculate current performance ratio as a comparison of the specific yield (kWh/kWp) of the PV system 3 compared to a set point based on irradiation data. The achieved performance ratio is then compared to a thresh- old based on geographic and climatic values. Interpretation directly as an alert "PC low" as well as a trigger for fur- ther analysis (degradation calculation). In a further step S5, the determined recommendation REC can be output by the recommendation engine 1B of the performance center cloud server 1 via the cloud 2 to a user interface or a customer interface UI of an operator U of the investigated photovoltaic system 3 and/or via a control interface to a controller of the monitored photovoltaic system 3. In re- sponse to the recommendation REC received by the controller of the monitored photovoltaic system 3, optionally, manual actuators provided at the respective photovoltaic system 3 can be activated and adjusted by the user U to increase the performance of the monitored photovoltaic system 3 according to the output recommendation REC. Additionally or alterna- tively, some actuators can be also controlled directly in re- sponse to recommendations received by the controller. For in- stance, a position and/or orientation and/or a tilt angle of photovoltaic modules or panels of the monitored photovoltaic system 3 can be changed automatically in response to the rec- ommendation REC received via a control interface from the recommendation engine 1B of the performance center cloud server 1. Further, in a possible implementation, the surfaces of the photovoltaic modules of the monitored photovoltaic system 3 can be cleaned manually by cleaning actuators han- dled by a user U in response to cleaning recommendations REC output via a display of a user interface UI to the user U or by cleaning actuators controlled directly in response to rec- ommendations REC received by the actuators from the recommen- dation engine 1B via the cloud 2. As already mentioned above, the cloud-based system as illus- trated schematically in Fig. 1 comprises a plurality of dif- ferent heterogeneous data sources providing performance rele- vant data which can be processed by the performance evalua- tion engine 1A and/or by the recommendation engine 1B of the performance center cloud server 1. The heterogeneous data sources can comprise internal data sources 4-i as illustrated schematically in Fig. 1 and/or external data sources 5-i as illustrated in Fig. 1. Further, one or more databases 6 can be used by the performance center cloud server 1 to perform performance evaluations and/or to calculate recommendations.

The internal data sources 4-i of the photovoltaic systems 3 can, for instance, comprise system configuration data indi- cating a configuration of different system components of the photovoltaic system 3-i. The system configuration data can be loaded in a possible embodiment from a database of the system of the respective photovoltaic system 3-i. Optionally, the system configuration data of different photovoltaic systems 3-i can be stored in a local database 6 of the performance center server 1. The system configuration data indicates the structure and topography of the photovoltaic system 3 and the interconnection between the plurality of different system components. Further, the system configuration data can com- prise data concerning local configuration of configurable system components such as system hardware components and/or system software modules. The system configuration data can additionally be stored in a local memory of the respective photovoltaic system 3-i.

The performance relevant data provided by internal data sources 4-i of the photovoltaic systems 3 can also comprise service messages (SM) and/or state codes (SC) generated by components of the monitored photovoltaic system 3 during ob- servation time periods.

Further, the performance relevant data provided by the inter- nal data sources 4-i of photovoltaic systems 3 can comprise flow data indicating a power generation and/or performance of the monitored photovoltaic systems 3 as time series data. The data may be processed, for example, as aggregated 5-minute values. Optionally, the performance relevant data provided by the data sources of the photovoltaic systems 3-i can comprise irradiation data indicating a solar radiation at the geo- graphic location of the monitored photovoltaic system 3-i. In case that irradiation sensors are installed at the location of the respective photovoltaic systems 3-1 the irradiation data can originate from internal data sources. Since in many use cases such irradiation sensors are not available at the location of the photovoltaic system 3-i the irradiation data is mostly provided from external data sources 5-i.

The performance relevant data provided by internal data sources 4-i of the photovoltaic system 3-i can comprise his- toric performance data related to a performance of a moni- tored photovoltaic system 3-i during observation time periods in the past.

Besides the internal data sources 4-i, the cloud-based system can also comprise external data sources 5-i as illustrated schematically in Fig. 1. The performance relevant data pro- vided by the external data sources 5-i can comprise, for in- stance, environmental data indicating environmental condi- tions at the location of the monitored photovoltaic system 3 over time. The environmental data can comprise in particular weather data and/or solar irradiation data of solar radiation hitting the photovoltaic modules of the photovoltaic system 3. Moreover, the environmental data can also comprise loca- tion specific data. The external data sources 5-i can also provide statistical data including failure and/or error prob- abilities of system components implemented in the respective photovoltaic system 3. The statistical data can also comprise failure or error one or more dimensional probability distri- butions. The statistical data provided by the external data sources 5-i can also comprise climate zone data related to the climate zone where the monitored photovoltaic system 3 is located.

The performance center cloud server 1 can process a plurality of different performance relevant data to provide a holistic view on the investigated photovoltaic system 3-i. Optionally, in the computer-implemented method the availabil- ity and/or connection status of the different internal and external heterogeneous data sources 4-i, 5-i is continuously monitored by the performance center cloud server 1 to assess or estimate a reliability and accuracy of the recommendations determined by the recommendation engine 1B. The internal data sources 4-i of the photovoltaic system 3 can be associated to different inverters of the monitored photovoltaic system 3 to provide inverter specific performance relevant data of the monitored photovoltaic system 3.

Fig. 3 schematically shows a cloud-based system for perform- ing the computer-implemented method for providing assistance for increasing a performance of a monitored photovoltaic sys- tem 3. The cloud server 1 illustrated schematically in Fig. 1 can provide different kinds of services to a user or operator of an investigated photovoltaic system 3. The performance evaluation engine 1A of the performance center cloud server 1 is adapted to provide a performance evaluation service to a user or operator of at least one photovoltaic system 3-i. Be- sides the performance evaluation service, the cloud server 1 is also adapted to provide an additional recommendation ser- vice for handling root causes having caused a degradation of a performance of an investigated photovoltaic system 3. The performance evaluation service can be provided by a perfor- mance evaluation engine 1A and the recommendation service can be provided by a recommendation engine 1B integrated in the same server 1 as illustrated schematically in Fig. 1. How- ever, alternatively, the performance evaluation service can be provided by a performance evaluation engine 1A being sepa- rated from a recommendation engine 1B within the cloud 2. In this example, the performance evaluation engine 1A and the recommendation engine 1B communicate via a data network or cloud 2 as illustrated in Fig. 1. For example, the perfor- mance evaluation engine 1A may perform steps S1, S2 of the computer-implemented method as illustrated in the flowchart of Fig. 2, whereas the recommendation engine 1B may perform steps S3, S4, S5 of the illustrated flowchart. To provide the performance evaluation service, the performance evaluation engine 1A is adapted to receive a plurality of performance relevant data from heterogeneous data sources such as a "so- lar,creator" and "solar.web". In the cloud-based system exem- plarily illustrated in Fig. 3, the performance evaluation en- gine 1A receives performance relevant data from the internal data sources 4-1, 4-2, 4-3. The data source 4-1 provides con- figuration data CD of the investigated photovoltaic system 3. The system configuration data CD can indicate a configuration of system components of the monitored photovoltaic system 3.

The performance evaluation engine 1A further receives from a second data source 4-2 "solar.web" different data streams. The internal data source 4-2 can provide service messages SM and/or state codes SC generated by components of the photo- voltaic system 3. In the illustrated example, the internal data source 4-2 can further provide historic performance data HD related to a performance of the monitored photovoltaic system 3 observed during previous observation time periods in the past. The third internal data source 4-3 performed by a state code software tool can provide interpretations of state codes SC of components of an investigated photovoltaic system 3. The state codes can be provided by inverters or data log- gers of the photovoltaic systems 3-i. As illustrated in Fig. 3, the system configuration data CD, the service messages SM, the flow data FD, the historic data HD as well as the inter- pretations of the state codes SC are supplied into the cloud 2 and to a performance evaluation engine 1A to provide a per- formance evaluation service. The data sources 4-1, 4-2, 4-3 include a plurality of internal data sources provided in a predefined group of photovoltaic systems 3-1 to 3-N. Besides the internal data sources 4-i, external data sources 5-i can be provided. In the illustrated embodiment, a first data source 5-1 is adapted to provide statistical data concerning error probabilities of common errors of predefined system components. A further external data source can comprise weather data concerning the local weather at the geographical location of the investigated photovoltaic system 3. A further database 5-3 can provide climate zone data related to a cli- mate zone where the investigated monitored photovoltaic sys- tem 3 is located. This data can also comprise Koppen-Geiger classification data using the Koppen-Geiger classification concerning climate zones. Moreover, the external data sources 5-i can comprise data relating to a partial shading 5-4 of photovoltaic modules of an investigated photovoltaic system 3. The external data sources 5-i can provide environmental data indicating environmental conditions at the location of a monitored photovoltaic system 3 over time. These environmen- tal data can comprise weather data read from the data source or database 5-2. Statistical data can be derived from the statistical database such as data source 5-1. The statistical data can include failure and/or error probabilities of system components of photovoltaic systems 3. These statistical data can also indicate failures and error probabilities of system components in relations to other measurable data such as tem- perature data or humidity data. Another data source for sta- tistical data is the database 5-3 providing climate zone data in relation to a climate zone where the photovoltaic system 3 can be implemented.

The performance center cloud server 1 can comprise a perfor- mance evaluation engine 1A adapted to perform different kinds of performance evaluation services and/or a recommendation engine 1B adapted to perform a recommendation service. In a possible implementation, the performance evaluation engine 1A is adapted to perform a neighborhood comparison service. This neighborhood comparison service includes screening and moni- toring of key performance indicators KPIs in relation to ref- erence photovoltaic systems located within a geographic prox- imity of the investigated photovoltaic system 3-i. One or more selected photovoltaic systems 3-i can be compared with each other with reference to their individual performance. For instance, a group of photovoltaic systems 3 within the vicinity of an investigated photovoltaic system 3 are se- lected automatically or manually by an operator of the dif- ferent photovoltaic systems 3 to calculate average reference performance benchmarks or key performance indicators KPIs which are used for performing a comparison with the corre- sponding monitored performance indicators of the investigated photovoltaic system 3-i. If, for instance, there are several photovoltaic systems 3 with an identical or similar configu- ration in the vicinity of an investigated photovoltaic system 3, the performance evaluation engine 1A can calculate average performance indicators on the basis of the data received from the data sources in relation to the reference photovoltaic systems 3 being located in the vicinity of the investigated photovoltaic system 3-i. The performance generated by the in- vestigated photovoltaic system 3-i is then compared in a pos- sible embodiment to the calculated average value for the other similar photovoltaic systems 3 in its vicinity to de- tect a possible degradation or underperformance of the inves- tigated photovoltaic system 3. In this embodiment, the refer- ence performance is provided by the calculated average per- formance of other photovoltaic systems 3 located in a prede- fined range around the geographic location of the investi- gated photovoltaic system 3. It is also possible to use a weighted average performance or a performance provided by ma- chine learning. Moreover, in a possible embodiment, the per- formance of an investigated photovoltaic system 3 within a predefined observation time period is compared with a refer- ence performance calculated on the basis of historic data HD concerning the performance of the same photovoltaic system 3 over a predefined observation time period in the past. The comparison result can be evaluated to determine long-term trends within the performance of an investigated photovoltaic system 3. The processing of the data can comprise machine learning methods. The calculation can be based on linear re- gression. Optionally, the performance evaluation engine 1A of the per- formance center server 1 can take into account one or more reference performances calculated on different databases. For instance, the evaluation engine 1A can take into account a relative performance of the investigated photovoltaic system 3 in relation to other similar photovoltaic systems 3 in its vicinity and also a relative performance of the investigated monitored photovoltaic system 3 at an observation time in re- lation to a performance provided by the same photovoltaic system 3 over a past observation time period.

Optionally, the performance center cloud server 1 is able to monitor a fleet of photovoltaic systems 3 of the same operat- ing entity. The photovoltaic system 3 is able to monitor and manage an entire photovoltaic system fleet which can be com- missioned in "solar.web" and monitored in "solar.web" or So- lar,web Business. When monitoring a plurality of photovoltaic systems 3 within a predefined group of photovoltaic systems 3, the performance center server 1 provides for the detection and interpretation of errors or degradations within the dif- ferent photovoltaic systems 3 at the different locations.

In the cloud-based system exemplarily illustrated in Fig. 3, the design software tool "solar.creator" allows an individual configuration of a photovoltaic system 3, wherein this con- figuration data CD can be considered for neighborhood compar- ison and/or for root cause analysis. The configuration gener- ated by the design tool "solar.creator" can be synchronized with the performance center cloud server 1. The recommenda- tion engine 1B of the performance center 1 offers, for exam- ple, an interpretation of health and performance key perfor- mance indicators KPIs based on the performance relevant data provided by the different data sources. The performance cen- ter 1 can interpret common errors (FLD files), weather data, location data and also partial shading detection data pro- vided at the different data sources 5-i. In this way, it is possible to interpret key performance indicators KPIs without single-string monitoring (I-V curve diagnosis).

As illustrated in Fig. 3, the interpretation of the key per- formance indicators (KPIs) can, for example, be supplied by the recommendation engine 1B providing the recommendation service and also to a photovoltaic system 3 detail dashboard DB of a customer interface. Further, the key performance in- dicators provided by the performance evaluation engine 1B can be supplied to a photovoltaic system status interface S1 of the customer interface as illustrated in Fig. 3. A detailed analysis can be performed on the basis of adapted real time charts AR-chart and/or adapted history charts AH-chartcon- cerning performance data of the investigated photovoltaic systems 3.

The performance center cloud server 1 can provide a general overview of the interdependencies of several components within the investigated photovoltaic system 3. The perfor- mance center cloud server 1 is adapted to calculate different kinds of key performance indicators KPIs based on the data provided by heterogeneous internal or external data sources. Optionally, the key performance indicators KPIs can also be displayed to various touchpoints and evaluated to derive rec- ommendations RECs for actions to be performed manually by a user or performed automatically by controllers at the inves- tigated photovoltaic system 3.

Optionally, the recommendation engine 1B can also comprise a machine learning module. This machine learning module can be used in a possible embodiment for a so-called RISO analysis performed by an inverter and/or to perform a voltage and power analysis in relation to the investigated photovoltaic system 3. Optionally, a connection status can be monitored as a key performance indicator KPI in the cloud-based system. For in- stance, the connection status can indicate a percentage of online data sources and/or online status of inverters availa- ble to the performance center cloud server 1. It may also in- dicate the number of available data sources within the dif- ferent photovoltaic systems 3. The different available data sources of the photovoltaic systems 3 can be displayed in a possible embodiment via a graphical user interface of a cus- tomer. In this way, a customer has the possibility to see how many of the available data sources are online and can provide data to the performance center cloud server 1. Further, log data values can be processed to form a basis for reports pro- vided to a customer such as an operator of a photovoltaic system 3. A customer can see in this way which data sources are unreliable and can take measures accordingly.

In a neighborhood comparison procedure, a daily curve of kil- owatt hour produced by the respective photovoltaic system 3 can be measured and displayed to a customer. Optionally, the overall photovoltaic system performance can compare the power at the DC side of an inverter to the power at its AC/DC side. Optionally, a customer can receive a comparison of a selected photovoltaic system 3 under investigation with average photo- voltaic systems of the same area to determine how the inves- tigated photovoltaic system 3 performs over time. A customer or user can also select specific photovoltaic systems 3 as reference systems for comparison with the performance of the investigated photovoltaic system 3.

Optionally, the performance center 1 allows for an inverter energy comparison. For example, the produced power may be in- vestigated and displayed on an inverter level. The electrical power output by an inverter can be displayed on a screen of the customer interface. In this way, it is possible to com- pare the performance of different inverters within the same photovoltaic system 3. In this way, a customer can see at a glance which strings (MPP level) within the investigated pho- tovoltaic system 3 have performed best within the observed time period in comparison to other strings of the same photo- voltaic system 3. In this way, a customer can also recognize easily a long-term degradation on MPP level. A deviation be- tween inverters can be displayed.

A further key performance indicator KPI provided by the per- formance center 1 can comprise a performance of a photovol- taic system 3 compared to a setpoint based on a measured so- lar irradiation. In this way, a customer can be informed about the performance of the investigated photovoltaic system 3 compared to an expected performance based on the solar ir- radiation at the corresponding location. In this way, the customer can determine whether the investigated photovoltaic system 3 is over-performing or under-performing taking into account the solar radiation at the location of the photovol- taic system 3. The performance ratio of an investigated pho- tovoltaic system 3 can be evaluated and displayed in a regu- lar time interval, e.g. a 15 minutes time interval, daily or for other observation time periods such as months, years or user-defined time frames.

A further source of information is formed by service messages SM generated by components of the photovoltaic system 3. The service messages SM generated by components can be archived for a predefined time period which depends on the photovol- taic system. The logged service messages SM can in a possible embodiment be preprocessed by classifying the different ser- vice messages SM. In a possible embodiment, the service mes- sages SM can be displayed to a customer via a customer inter- face. In a possible embodiment, the customer has the possi- bility to sort the service messages SM according to prede- fined criteria for instance chronologically by a classified severity and/or by a state code SC and/or according to the corresponding inverter within the investigated photovoltaic system 3. A pattern recognition is performed on the basis of the service messages SM.

The performance center cloud server 1 can, for example, also perform a long-term degradation analysis. For instance, the performance evaluation engine 1A is adapted to detect a deg- radation of the insulation resistance of the photovoltaic system 3 which may cause later start-ups of the photovoltaic system 3 and may cause a shutdown of the photovoltaic system 3 on rainy days thus affecting the performance of the respec- tive photovoltaic system 3. Further other failures caused for instance by bad cable connections or by broken cables can be detected.

Optionally, the performance center cloud server 1 can also detect a long-term degradation by performing a voltage analy- sis. An interpretation of the voltage and/or performance deg- radation can be performed. For instance, a customer can be informed by means of the customer interface about a long-term degradation of the performance and voltage of an investigated photovoltaic system 3 and can also get an interpretation or recommendation of a requested action. The recommendation can for instance comprise a hint that a photovoltaic module of the photovoltaic system 3 might be shaded or dirty or PID suspicious. In a further long-term investigation, the perfor- mance center cloud server 1 may calculate a trend of the in- vestigated photovoltaic system 3 in comparison with other reference photovoltaic systems. In a possible embodiment, a customer can see via its customer interface how the investi- gated photovoltaic system 3 is performing on the long run when compared to the average similar photovoltaic system 3 in its geographic proximity.

Optionally, the performance center 1 provides a real time view of one or more selected key performance indicators KPIs. The real time values can be calculated event-driven, for in- stance when the user opens a specific view via a graphical user interface of its customer interface.

The output time series values can be inverter specific values of inverters within the investigated photovoltaic system 3. For example, a serial number or another identifier of the re- spective inverter within the investigated photovoltaic system 3 can be displayed via a screen of the customer interface. Different data sources of the investigated photovoltaic sys- tem 3 as well as inverters can be sortable and may be fil- tered by using specific filters adjusted by the customer in- terface. In this way, it is clearly visible to a customer or user which inverter belongs to which data source. Addition- ally, a total power or a power per MPP can be displayed to a user via a customer interface of the cloud-based system as illustrated in Fig. 3.

The computer-implemented method according to the present in- vention allows to specify different photovoltaic systems 3 in geographic proximity that can be compared with each other and can track their specific yield on a module field level. The mapping of module fields and inverters can be performed to provide comparability of different values. Based on that, photovoltaic module fields can be compared with each other based on their specific yield (kWh/kWp). For performing a comparison, for example, three different types of photovol- taic systems or photovoltaic module fields can be distin- guished. These types comprise a photovoltaic system 3, i.e. a photovoltaic system in the neighborhood of an investigated photovoltaic system to perform a comparison of the measured key performance indicators KPIs. Further, the photovoltaic systems can comprise so-called considered photovoltaic sys- tems, i.e. all photovoltaic systems that are qualified for a comparison solely based on their geographic location and con- figuration (orientation, tilt angle etc.). Further, the pho- tovoltaic system 3 can also comprise selectable photovoltaic systems from a subset of the considered photovoltaic systems where a specific account of a user has access rights on. From the considered photovoltaic systems, only average values can be displayed to the customers via the customer interface among the selectable photovoltaic systems 3. A customer or user can be able to freely select several reference photovol- taic systems for the benchmark comparison and may save favor- ites among them.

The cloud-based system as illustrated in Fig. 1 can also cal- culate a performance ratio as a basis for performance guaran- tee reports. A performance guarantee report can support an installer of the photovoltaic system 3 in offering perfor- mance guarantees. These performance guarantees can for in- stance guarantee a certain yield that the respective photo- voltaic system 3 is required to generate and possibly supply to a connected public power supply grid or for self-consump- tion. The value can be calculated in a possible embodiment in different time periods, for instance each day or each month or year.

Optionally, the photovoltaic system 3 can be configured by defining a structure or topology of the respective photovol- taic system 3 and by indicating the types of the used photo- voltaic components. For example, a photovoltaic system 3 can consist of a predefined number of photovoltaic module fields. Each photovoltaic field can comprise a predefined number of photovoltaic modules which are collected to strings. These strings can itself be connected with MPP trackers of an in- verter. Each inverter of the photovoltaic system 3 can com- prise one to a predefined maximum number of MPP trackers de- pending on the inverter model. For performance guarantee re- ports, an STC (Standard Test Conditions) performance ratio can be calculated according to IEC61724-1. The performance center 1 can also be used to perform a volt- age and power analysis. In a possible implementation, an ade- quate filtering method is employed so that the analysis re- sult gets less dependent on environmental influences, in par- ticular weather conditions. For example, a subroutine can be performed which is adapted to detect the local weather condi- tions based on local available data provided by local sensors or local inverters. This can be important if an access to re- mote databases providing weather data is not possible. For instance, weather clouds passing over a photovoltaic system 3 can cause a variation of the generated electrical power over time so that a corresponding signal reflecting the power var- iation can be generated. For instance, if the amplitude of the generated signal is varying over time, a frequency of the variations can be determined or calculated. This frequency reflects the clouds passing over the investigated photovol- taic system 3. For instance, if a cloud is passing over the photovoltaic modules of the investigated photovoltaic system 3, every minute there is a power generation variation within the detected signal reflecting this passage rate of the clouds. The power measurements of the inverter reflect the passage of the clouds. The amplitude of the measured power of the photovoltaic system 3 is reduced from a high power ampli- tude to a low power production amplitude whenever a sun-shad- ing cloud is passing over the photovoltaic array of the re- spective photovoltaic system. In this way, it is possible to perform a clear sky detection to classify if the rate of the sun-shading clouds is low, e.g. when the frequency of the am- plitude changes is low. In this implementation, the frequency of the detected weather clouds passing over the photovoltaic system 3 can form a further performance relevant data source which can be taken into account by the performance evaluation engine 1A of the performance center cloud server 1.

The recommendation engine 1B can be adapted to perform a rec- ommendation service, i.e. provide recommendations for per- forming possible actions in relation to components of the in- vestigated photovoltaic system 3. Recommendations RECs can be displayed on a screen of the customer interface so that an operator or user can perform necessary actions to increase the performance of the photovoltaic system 3. Further, recom- mendations RECs can also comprise control flags and/or con- trol signals supplied via the cloud 2 to one or more control- lers of the investigated photovoltaic system 3 to trigger di- rectly or indirectly adjustments at system components of the investigated photovoltaic system 3 which are suitable to in- crease the performance of the investigated photovoltaic sys- tem 3 in view of the performance evaluation and/or detected root causes DRCs.

The performance center 1 can generate automatically alerts which are displayed on a screen of the customer interface in case that specific parameters or calculated key performance indicators KPIs require immediate action at the investigated photovoltaic system 3. The customer interface as illustrated in Fig. 3 can be provided by a web-based application or can be integrated in a portable user equipment carried by a user or operator of a photovoltaic system 3 or a photovoltaic sys- tem fleet.