The invention relates to a system and method for monitoring and/or diagnosing operation of a production line of an industrial plant, wherein the operation of the production line is controlled by an automation system and wherein at least a first and a second operational state of the production line are dependent on at least one control input signal to the production line. .

Document EP 2 293 164 A1 discloses a process control and monitoring system for monitoring a distributed primary system, such an electric power system, a water or gas supply system or a telecommunication system. The process control and monitoring system is arranged in a distributed computing resource, called cloud resources. The process control and monitoring system receives via network connection sensor signals of the primary system and provides it to the user via a network connection as well.

It is an object of the invention to improve supervision and maintenance of a production system comprising a production line and an automation system controlling the production line.

The object is solved by the features of the independent claims. Advantageous embodiments are subject of the dependent claims.

According to an aspect of the invention, the system for monitoring and/or diagnosing operation of a production line of an industrial plant comprises a remote data processing server installed outside of the industrial plant which is connected to the production line and/or automation system via a first data communication network and which contains an output interface either to a monitoring and/or diagnosing module implemented in the remote data processing server or to the first or a second data communication network (214), wherein the remote data processing server is arranged to

receive over the first data communication network a digital input signal reflecting the at least one control input signal, and a digital output signal reflecting the second operational state,

determine at least a first and a second modeled state corresponding to the at least first and second operational states, respectively, by inputting the digital input and the digital output signals to a digital observer model modeling the time-dependent operational behavior of the production line and the automation system and by processing the digital observer model, and

forward the at least first and second modeled states to the output interface.

According to another aspect of the invention, the method for monitoring and/or diagnosing operation of the production line comprise the method steps performed by the remote data processing server. The production line which is to be monitor and/or diagnosed might exemplary include as actuators robots, manipulators, drives and/or pumps that are for example interconnected by conveyors and/or pipes. The actuators are part of a real industrial plant that receives raw materials, e.g from tanks, and converts them into products based on a production process that is realized by the production line controlled by the automation system, where the production process is defined based on predetermined conditions e.g. by an operator. The actuators are provided by the automation system with the at least one control input signal, which causes the actuators to perform resulting control actions influencing the behavior of the production line and thereby the way the production process is executed. In the following, the combination of production line and automation system is also called production system.

The digital input signal which reflects or contains information on the at least one control input signal has preferably a value which corresponds to the magnitude of the relating control input signal. The digital input signal may reflect the at least one control input signal as it is currently applied to the production line, which is preferable for online monitoring and diagnosing. For offline monitoring and diagnosing, historical control input signals may be sent to the remote data processing server.

The provided system and method are based on the thought to reduce the number and amount of technical equipment needed in an industrial plant for providing required monitoring and/or diagnosing signals and information.

This is achieved by providing a digital observer model which models the time-dependent behavior of the production line and automation system, i.e. it consists of a production line model and an automation system model, and which can be used to calculate operational states of the production line model and of the automation system model when the control input signals and at least one of the real operational states are known. Compared to the currently installed, plant-based monitoring and diagnosing systems, the number of sensor signals required and therefore of sensor units to be installed are reduced considerably, as many of the operational states are now calculated rather than measured. Further, it may now become possible to determine internal operational states which cannot be obtained through measurement by generating them with the digital observer model.

In addition, the processing of the digital observer model is implemented in a remote data processing server which is installed outside of the industrial plant. Accordingly,

the amount of data processing capacity which is required directly in the industrial plant for monitoring and/or diagnosing purposes is reduced as well.

The determined modeled states of the digital observer model are forwarded in one solution to a monitoring and/or diagnosing module implemented in the remote data processing server. The monitoring and/or diagnosing module may then process the modeled states as if they were measurements of the real production system and may then represent the outcome and results to an operator of the industrial plant via a Web page, for example. This means that the industrial plant, in addition to the few sensor signals, only needs to provide access to the one software application which runs on the remote data processing server for visualizing the monitoring and diagnostics results. The access to this software application is easily realized also from mobile communication devices, so that not only operators sitting in a control room, but also service personnel working in the field, can simultaneously use the digital observer model for monitoring and/or diagnosing purposes.

In an alternative solution, part of the monitoring and/or diagnosing functionality is

implemented on the side of the industrial plant. In that case, the remote data processing server is arranged to pre-process the at least first and second modeled states for the monitoring and/or diagnosing purposes before forwarding them to the output interface.

The provided system and method may in addition be used to facilitate an efficient control of the production line. Therefore, the digital observer model may serve for simulating the geometric, the kinematic and/or the behavioral conditions of the production line or parts of it, as exemplary production systems or control components or devices in the real plant object. The digital observer model may comprise models of devices, processes and signals of the production line, copies and or simplified models of the control functions performed in the automation system and also models of an interaction between the production line and the automation system. The digital observer model thus represents an imitation of the time- dependent behavior of the real plant.

However, the simulation of the real plant , i.e. the processing of the digital observer model requires suited simulators, emulators and stimulators that calculate an inner time-dependent state of the digital observer model, without having measurements of the corresponding real components of the real plant present. The simulation can therefore be used not only to determine the current operational states of the production system based on just the control input signals and a few sensor signals, but also to predict a future time-dependent behavior of the real plant. In addition, the digital observer model may be used to optimize the production process based on the results of the digital observer model, to validate the functionality or performance of elements of the production line or of control functions of the automation system, to train operators of the industrial plant, to schedule the production of the production line and so on. Dependent on the required quality and precision of the modeled states, the digital observer model may be modeled in different abstraction levels.

In line with the provided system and method it is recognized, that the behavior of the production system is predicted by calculating the modeled states as discrete time-dependent signals, whereby it is assumed that the real time-dependent behavior of the operational states of the production system and the time-dependent behavior of the modeled states is the same.

As this can usually not be ensured during the whole running time of a production process, it is suggested that the digital input signal and the digital output signal each contain a time stamp indicating a point in time when the at least one control input signal and when the second operation state were taken, respectively, and wherein the remote data processing server is arranged to time synchronize the processing of the digital observer model with the time stamps of the digital input and output signals.

Nevertheless, changes during the operation of the real plant, such as services and maintenances are not fed back to the digital observer model, such that a difference between the operational and the modeled states may steadily increase, and the reality is not reflected any more.

Therefore, in line with the provided system and method, it is proposed that the remote data processing server is arranged to compare the digital output signal with the second modeled state and to adapt the digital observer model based on the result of the comparison. It is further proposed to start the processing of the digital observer model from a well-defined initial state, where the initial state may for example be achieved by starting the production process and the processing of the digital observer model simultaneously, or by waiting until the production system have reached a steady state or a well-defined state and by triggering the processing of the digital observer model accordingly.

It is further suggested to collect data in the production system that describe a specific operational state of the production system, when the automation process in the real plant already runs. This proposal is based on the thought that the initial state of the production system and therefore of the digital observer model is unknown when starting to simulate the behavior of the real plant, where this may lead to the above mentioned difference between the simulation and the reality. Based on that thought, an information interface to the automation system in the shape of a data collector is provided which has access to all required information that describe an initial or well-defined state of the production system. This data collector system may exemplary be distributed in different sub components responsible for collecting information from different parts of the production system. Once a certain operational state of the running production system is known, it might be used to update the digital observer model to close a gap between the simulation and the reality. In other words, the provided method proposes to track the digital observer model to the reality at least from time to time.

Starting from the tracked digital observer model, the simulation of the real plant is closer to the reality such that exemplary break downs can be estimated with a higher reliability. Even if a breakdown or another error could not be timely detected, the tracked digital observer model can be used to reconstruct the situation leading to the breakdown or error in a more reliable way, since in contrast to an initial boundary condition used to start the simulation of the real plant based on the digital observer model, the last operational state might already include an information about a trigger that leads to the breakdown or the error. Consequently, a downtime of at least parts of the real plant can be reduced. In this respect, the proposed method also reduces the hardware effort to track the trigger leading to the breakdown or the error, since all necessary data to analyze the broken down or erroneous part of the real plant are already available in the digital observer model and need not to be tracked in the field in a cumbersome way. A special service team in the field for tracking data in the field with measurement instruments will also become at least partly superfluous. As production lines and production systems usually are comparatively complex, it is further suggested that the digital observer model is divided into at least two sub-models modeling the time-dependent operational behavior of at least two corresponding parts of the production line and the automation system.

Further, in line with the provided system and method, the operational states of the production system may comprise all information that is necessary to fully describe the characteristics of the production system at or up to a certain point in time. However, in line with the provided system and method, the operational states of the production system might also only comprise a part of this information, depending on the monitoring and/or diagnosing requirements. The operational states can be for example include signal values of sensors or actuators, software states, information about internal variables or instruction pointers of the software programs running on control devices and communication systems, i.e. belonging to the automation system.

The proposed system and method enable a service team to analyze the production system's behavior or to predict maintenance. However, in line with the provided system and method, it is recognized that with the formerly used techniques, access to the operational states of the production system was limited, as the service team had no view of those signals which were not or could not be measured. This view is now provided by the digital observer model. In this way, the service team is able to check all necessary information about the production system based on the digital observer model which modeled states are sufficiently close to the real operational states of the production system.

In a further embodiment, the remote data processing server is implemented on a distributed processing system including distributed storage means. Accordingly, the provided method comprises the step running the digital observer model on a distributed processing system including distributed storage means, called cloud, where the distributed processing system is connected to the production system via the first network connection. In line with the provided system and method, a cloud shall be a network of computers that behave like a single computing component. It might be used for a wide range of applications from low complex applications, e.g. storage, up to high complex applications which are able to provide complex and full featured virtual PCs with their specific software, interfaces and operating system. A simple cloud application might be a shared storage that behaves like an external hard drive, but provides synchronization functionality that mirrors data from the cloud to a local folder. Changes in both the cloud and the local folder are automatically synchronized in the background without user interaction. Different vendors provide storage that allows storage of data in a shared storage that is permanently synchronized with a local folder on the personal computer of different users. For this, the vendor specific software is installed on each individual PC that permanently performs the synchronization process in the background in case of online access. Furthermore, this technology allows file sharing across different operating systems. Synchronization software is provided for several platforms, like MS Windows, Android, Linux, etc. that allows data synchronization and updating from mobile device, as well. Finally, versioning functionality on file level is provided, which allows restoring data from older versions and tracking changes.

A complex cloud application, as in the present embodiment, might be the utilization of a network of computers used to share computing power between several different types of servers, like data servers, active directory servers, etc., offered for executing functions and algorithms directly in the cloud. The advantage of the cloud technology is that not only storage, but also computing power may be temporarily or statically ordered, in order to perform larger software on a specific virtual machine or move more virtual machines into it. Dependent on the complexity of the digital observer model, the cloud technology can thus be used to run it on distributed computation resources to achieve the simulation results before the modeled states of the digital observer model become out of date, since the operational states of the real production system already moved further. In a further embodiment the provided method comprises the step storing the updated modeled states in the distributed storage. This allows not only accessing the modeled states as long as a network connection to the distributed storage is available, which clearly facilitates the remote monitoring and diagnosing of the real production system, the data traffic of the updated modeled states might also become very high, which might be

impossible to be stored on a storage device locally provided at the production system. In this respect, the distributed storage and thus the cloud can enable to reduce occurring data traffic and to store all versions of the updated modeled states over a predetermined time period, to give a service team insight into the operational states of the production system at any later time based on the digital observer model, without any interruption over this predetermined time period. In an embodiment of the provided system and method, the at least first and second modeled states are determined continuously, periodically or on a triggered basis. While a continuous updating of the modeled states always provides information on the most actual operational states of the production system, a periodical update reduces the required data traffic. When using a triggered updating, the updating can be controlled, for example to achieve the best results in case of a given limited bandwidth to transmit the measured operational state of the production system. In another embodiment of the provided system and method, the measured operational state or states of the production system includes at least one of

- signal values of sensors and/or actuators of the production system, and/or

- software states of the automation system, and/or

- information about internal variables and/or instruction pointers of functions or software applications currently executed on devices belonging to the production system

and/communication systems communicating with the automation system.

In a further embodiment of the provided system and method, the modeled states of are determined with a speed that is higher than a speed in which the operational states of the production system are changed. By that means, it is possible to predict a future behavior of the production system and to recognize possibly occurring errors or breakdowns early enough to prevent them.

In yet another embodiment, , if there is a change required in the production system, this change can be tested beforehand by using the digital observer model and starting from a current state of the production system. The digital observer model may include abrasion and other time dependent influences on at least parts of the production system. Thus, the effect of the required change on the production system can be tested based on a simulation with the digital observer model with a higher reliability.

One further advantage of the provided system and method is that a bottleneck of the production system can by identified based on the running model. Synchronously thereto, scenarios for preventing the bottleneck can be tested in advance without affecting the operation of the production system. According to another aspect of the invention, the remote data processing server is arranged to process a training model containing the digital observer model and a model of a control room system, wherein the control room system includes an input for an operator to affect operation of the automation system.

Further, a method for training an operator of an automation system comprises:

performing an above provided method;

storing the updated modeled states; and

starting a training model to simulate the production system based on the updated modeled states, wherein the training model includes an input for the operator to change the updated modeled states.

The stored updated modeled states may reflect unforeseeable situations for the production system, which are difficult to establish merely based on mathematics. These unforeseeable situations can thus be suitably used to train an operator of the production system to react suitably.

In an embodiment of the invention, the remote data processing server includes a memory and a data processor. Therein, one of the above mentioned methods is stored in the memory as a computer program and the processor is adapted to carry out the method steps of the method, when the computer program is loaded from the memory into the processor.

According to a further aspect of the invention, a computer program is adapted to perform one of the above provided methods.

According to a further aspect of the invention, an electronic storage means stores a provided computer program and/or a digital observer model that is updated based on one of the above provided methods. The above described characteristics, features and advantages of this invention as well as the way, in which they will be reached, will be more clearly defined based on the following description of the embodiments of the invention, that will be described based on the figures, wherein Fig. 1 shows a schematic of a production system that is connected to a model of the production system in a cloud-based storage system,

Fig. 2 shows an example for the production system of Fig. 1 , Fig. 3 shows a schematic signal flow of the production system of Fig. 2,

Fig. 4 shows a schematic signal flow of a simulation program simulating the production system of Fig. 2, Fig. 5 shows a diagram illustrating the time-dependent behavior of one state variable in the digital observer model of the production system according to Fig. 4, and

Fig. 6 shows a schematic overview of the interaction between a production system and a remote data processing server.

In the figures, equal technical elements will be provided with equal reference signs and described only one time.

Reference is taken to Fig. 1 that shows a schematic of a production system 2 that is connected via a wire-bound or wire-less first data communication network to a digital observer model 4 of the production system 2 in a cloud-based remote data processing server, in the following simply called cloud 6.

The production system 2 includes a production line 8 that exemplary receives raw materials and forms a product out of these raw materials by performing a production process. This production line 8 and its production process is controlled by an automation system 10, also called a control system, that may provide control input signals 12 to the production line 8, and in particular to actuators in the production line 8, and receive measurement values 14 of operational states of the production line 8. In the present embodiment, the automation system 10 shall comprise at least one programmable logic controller, called PLC 10 hereinafter. Amongst others, the PLC 10 may exemplary constitute a closed control loop based on these values 12, 14 by controlling the production line 8 based on the control values or control input signals 12 that are itself generated based on the measurement values 14. In the present embodiment, a data collector 16 is provided that collects data 18, in particular a digital input signal reflecting at least one of the control input signals, and a digital output signal reflecting at least one of the operational states, from the production system 2 to transmit these data 18 to the digital observer model 4 in the cloud 6. The part of data 18 which comes from the production line 8 and from the corresponding production process can exemplary comprise state variables 20, sensor variables 22 or actuator variables 24. The part of data 18 which comes from the PLC 10 can exemplary comprise control pointers 26, values of internal variables 28 or software states 30. The data 18 might be exemplary collected via a data bus 32 and thus provided to the data collector 16.

In the cloud 6, the digital observer model 4 may include a process model 34 and a PLC model 36 to simulate the time-dependent behavior of the production system 2, wherein process model 34 represents both the production line 8 and the production process performed by it. The modeling of systems as the production system 2 is well known to the skilled person in the art and should thus not be explained in detail for the sake of brevity. In line with the production system 2, the process model 34 is controlled by the PLC model 36 that may also constitute a closed control loop based on respective control values 12' and modeled states 14' that result from the before mentioned simulation of the production system 2.

The simulation itself is performed by a simulation program 38 that utilizes the models 34, 36 to simulate the time-dependent behavior of the production system. Further, an analyzer component 40 may be provided that enables a service personell to look into the models 34, 36 and to analyze the simulated production system 2.

To explain the embodiment in further detail, an exemplary production system 2 should be regarded hereinafter. This exemplary production system 2 is shown in Fig. 2.

The exemplary production system 2 shall transport via a conveyor belt 42 metal sheets 44 that are collected in a container 46 to form a lamination stack 48. Such a lamination stack 48 might be used as rotor for an electromotor in a way well known to the skilled person in the art. The collected metal sheets 44 forming the lamination stack 48 are connected together via a mechanical press 50 in a way well known to the skilled person in the art. The conveyor belt 42 includes an electric conveyor motor 52 that is velocity controlled by the PLC 10. For that purpose, the PLC 10 outputs a conveyor current 54 to the conveyor motor 52 based on a received conveyor velocity 56 from the conveyor motor 52. In operation, the metal sheets 44 might be loaded on the conveyor belt 42 exemplary by an assistant. The conveyor belt 42 controlled by the PLC 10 in the way mentioned before transports the metal sheets 44 to the container 46 that will thus be filled with the sheets. An electric eye 58 detects the metal sheets 44 on the conveyor belt 42 and outputs for every passing metal sheet 44 an impulse 60 to the PLC 10 such that the PLC 10 can count how many metal sheets 44 are in the container 46. If there are enough metal sheets 44 in the container 46, the PLC 10 stops the conveyor belt 42 by stopping the conveyor current 54 and starts operating the mechanical press 50. The mechanical press 50 is driven by an electric press motor 62 that is force controlled by the PLC 10. For that purpose, the PLC 10 outputs a press current 64 to the press motor 62 based on a received press force 66 from the press motor 62. After the operation of the mechanical press 50, when the metal sheets 44 are connected together, the PLC 10 stops the production process of production line 8 until the container 46 is unloaded.

In the present embodiment, the above mentioned assistant might unload the produced lamination stack 48 from the container 46 and continue the operation of the production line 8 by pressing a start button 68 that outputs a start signal 70 to the PLC 10.

That is, in the present embodiment the production line 8 of the production system 2 includes two actuators 52, 62, which are synchronized in their operation by the PLC 10.

Reference is taken to Fig. 3 that shows a signal flow plan of the production system 2 of Fig. 2.

As already mentioned, the PLC 10 of Fig. 2 provides two closed control loops. In the first of these closed control loops, the PLC 10 controls the conveyor motor 52 with a conveyor controller 72. In the second of these closed control loop, the PLC 10 controls the press motor 62 with a press controller 74.

In the first closed loop, the PLC 10 generates a not referenced difference between a desired velocity 76 that is stored in a velocity storage 77 and the conveyor velocity 56 in a way known by the skilled person in the art. The conveyor controller 72 receives that difference and generates the conveyor current 54 in a way reducing the difference to zero. Any conveyor disturbances 78 on the conveyor motor 52 might increase the before mentioned difference between the desired velocity 76 and the conveyor velocity 56 and thus lead to a change in the conveyor current 54. Such a conveyor disturbance 78 might exemplary be fouling or wear in the conveyor belt 42 requiring conveyor motor 52 to drive the conveyor belt 42 with a higher force to keep the desired velocity 76. This higher force requires a higher conveyor current 54 to drive the conveyor motor 52. Consequently, any change in the conveyor current 54 might indicate a conveyor disturbance 78.

In the second closed loop, the PLC 10 generates a not referenced difference between a desired force 82, that is stored in a force storage 83, and the press force 66 in a way known by the skilled person in the art. The press controller 74 receives that difference and generates the press current 64 in a way reducing the before mentioned difference to zero. As in the first closed loop, any press disturbance 80 in the second closed loop might increase the before mentioned difference between the desired force 82 and the press force 66 and thus lead to a change in the press current 64. Such a press disturbance 80 might exemplary be a decreased lubrication of a gear in the mechanical press 50 requiring the press motor 62 to drive the mechanical press 50 with a higher force to keep the desired force 82. As in the first closed control loop, the higher force requires a higher press current 64 to drive the press motor 62. Consequently, any change in the press current 64 might indicate a press disturbance 80.

To run the production line 8 in Fig. 2, the PLC 10 has to synchronize the operation of the conveyor controller 72 and the press controller 74. As mentioned below, the conveyor belt 42 shall start to run when the assistant presses the button 68 providing the start signal 70 and as long as the amount of metal sheets 44 in the container 46 does not exceed a certain value 84 that might be stored in a limit storage 85. For that purpose, the PLC 10 includes a counter 86 that counts the received impulses 60. When the amount of received impulses 60 exceeds the certain value 84, the counter 86 outputs a counter signal 88. A comparator 90 compares the start signal 70 and the counter signal 88 and outputs a comparator signal 92 that activates the conveyor controller 72 and thus the conveyor motor 52, when the start signal 70 is active and the counter signal 88 is inactive. If respectively, the amount of received impulse signals 60 in the counter 86 gets equal to the certain value 84, the counter signal 88 of the counter 86 gets high, activating the press controller 74 and respectively deactivating the conveyor controller 72 via the thus deactivated comparator signal 92. In the present embodiment, the data collector 16 shall exemplary collect as data 18 all digital input signals reflecting the control input signals generated by PLC 10 and transmitted to actuators in production line 8, as well as at least one operational state of the production line 8 and/or at least one operational state of the PLC10. These collected data 18 are then transmitted to a storage location in the cloud 6, which storage location is accessed by the simulation program 38.

Reference is taken to Fig. 4 that shows a signal flow plan of the simulation program 38 that simulates the production system of Fig. 2 based on the process model 34 and the PLC model 36. In the models 34, 36 of Fig. 4, all simulated elements that correspond to real elements in the production system are referenced with the same reference sign as in Figs. 2 and 3. Therein, the simulated element is indicated by an apostrophe and will not be described one more time for the sake of brevity.

Based on the simulation program 38 and the used models 34, 36 therein, the production system can be monitored and/or diagnosed in detail. Amongst others, such a diagnosis might be directed to the question whether the operation of the two actuators 52, 62 in the production system 2 is well synchronized, or whether bottle necks occur in the provision of raw material. Another possibility might be a simulation of possible changes in the production system, wherein amendments in the production rate might be assessed by a systematic sequence of automated tests. The monitoring might further include finding potential error cases, which might lead to a breakdown of a part of or even the complete production system 2. Alternatively or additionally, the above mentioned assistant or other service personell might be trained based on the simulation program 38 prior to working with the real production system 2.

For the above mentioned purposes, the process model 34 and the PLC model 36 might be set to initial start conditions, called boundary conditions, when a simulation of the production system 2 should be started. Then, the simulation of the production system 2 might be run based on the process model 34 and the PLC model 36 until a point in time, where a time period of special interest for the above mentioned monitoring and/or diagnosing starts.

Exemplary, in case of an error, the simulation of the production system 2 might be run until a point in time shortly before the error occurred.

However, there may be situations where the chance to find an error or the reason for that error is small, because the process model 34 may not consider all possible side effects, such as aging or wear and tear of elements of the production line 8. Examples for un-modeled behavior of production line elements are the conveyor disturbance 78, such as fouling in the conveyor belt 42, and the press disturbance 80, such as the decreased lubrication of a gear in the mechanical press 50, i.e. these disturbances 78 and 80 are unknown for the process model 34.

This drawback can make it hard to find an error or at least the reason of the error based on the before mentioned simulation of the production system 2. In such cases, the field work of a service engineer at the real production system 2 would still be required. However, such a service engineer must travel to the real production system 2 and is not available for other purposes like testing optimization scenarios or providing remote assistance to other customers. Thus, the before mentioned error causes considerable extra effort in two aspects, namely due to the downtime of the real production system when the service engineer travels to the real production system and due to unavailability of the service engineer at his home place. In other words, the effort for the maintenance of the real production system 2 might be reduced considerably, when the simulation program 38 with the program model 34 and the PLC model 36, together with data of at least one operational state of the production system reflecting a subsequent effect of the error, can be used to identify the above mentioned error. To enable diagnosis for finding the before mentioned error or other kinds of the above mentioned monitoring or tests, it is proposed in line with the present embodiment to regularly re-align the process model 34 and the PLC model 36 with the data 18 that are collected by the data collector 18 in the production system 2. For that purpose, the simulation program 38 in the cloud 6 comprises a data distributor 94 that receives the data 18 from the data collector 16 in the real production system 2 and distributes the data 18 to signals in the simulation program 38, which correspond to the signals in the real production process 2 from which the data 18 has been taken. This distribution might accordingly be performed via a virtual data bus 32'. In the present embodiment, the transmission of the data 18 and thus the re-alignment between the real production system 2 and the simulation program 38 should be exemplary triggered based on the start button 68. That is, every time the assistant presses the start button 68, the process model 34 and the PLC model 36 will be updated based on the data 18 transmitted from the production line 8 and the PLC 10 of the production system 2. By that means, each production cycle in the automation system 2 following on the re-alignment can be simulated closer to the reality, which is particularly useful if an error occurs during the production cycle following on the synchronization.

In case of an error, the above mentioned service engineer can access the simulation program 38 exemplary in the cloud 6 via the analyzer component 40, which can be any computer device, in particular a handheld computer or tablet PC, on which a monitoring and/or diagnosing software application is running. The service engineer can thus reconstruct the real situation in the production system 2 based on running the simulation program 38 with the process model 34 and the PLC model 36. The service engineer can debug the sequence of the program steps in the PLC model 36 for exemplary finding any missynchronization between the conveyor belt 42 and the mechanical press 50. He can also supervise some or all signals in the simulation program 38 or even test a reaction of amendments in the production system by resetting some of the values in the simulated storages 77', 83'. For a clearer representation of the facts, in Fig. 4 the analyzer component 40 is connected to only a part of the signals, storages and other elements in the models 34, 36. However, principally, the service engineer shall get access to all elements in the models 34, 36 that are of interest to him.

The simulation program 38 might exemplary include the possibility to store all or a part of the operational states of the models 34, 36 in a database 96, which preferably is located in cloud 6 as well. By that means, digital copies of the operational states of the real production system 2 that might be of interest for a further analysis, for an above mentioned training of the above mentioned assistant or for any other suitable purpose can be stored in the database 96 and can be retrieved exemplary by the analyzer component 40 at any suitable point in time. The database 96 thus allows storing a history of the real production system 2 to increase the available information about the real production system 2 in the cloud 6.

Reference is taken to Fig. 5 that illustrates in a diagram showing a current 98 over time 100 the time-dependent behavior of the real conveyor current 54 and the simulated conveyor current 54'.

When the simulation program 38 starts at a first point 102 in time 100 and when the initial start conditions, i.e. the boundary conditions, of the production system 2 and of the simulation program 38 are the same, the real conveyor current 54 and the simulated conveyor current 54' are equal. Nevertheless, due to the above mentioned conveyor disturbance 78, the real conveyor current 54 and the simulated conveyor current 54' start to drift away from each other such that the conveyor current 54 will be simulated with a simulation error 104 at a second point 106 in time 100. At the second point 106 in time 100, the simulation error 104 is compensated by the above mentioned re-alignment of operational states between the automation system 2 and the models 34, 36 in the simulation program 38, such that the real conveyor current 54 and the simulated conveyor current 54' are set equal at the second point 106 in time 100. In the present embodiment, this method will be repeated at a third point 108 in time 100, a fourth point 1 10 in time 100 and a fifth point 1 12 in time 100. As already mentioned, this realignment at these points 102, 104 to 1 12 in time 100 might be triggered by pressing the start button 68. As a result, the simulation error 104 will be kept smaller than a further simulation error 1 14 that results, when the conveyor current 54 will be simulated without any re-alignment with the automation system 2 that is indicated in Fig. 5 with a pointed line provided with the reference sign 54". In Fig. 6, a schematic overview of the interaction between a production system 220 of an industrial plant and a remote data processing server 200 is shown, where the remote data processing server 220 is installed outside of the industrial plant. Analogously to what is described above, the production system 220 comprises a production line 223 and an automation system 226 for controlling operation of production line 223. The production system 220 is connected to the remote data processing server 200 via a first data

communication network 228. The remote data processing server 200 is arranged to process a digital observer model 202 which models the time-dependent operational behavior of the production line 223 and the automation system 226. Further, the remote data processing server 200 contains an output interface 203 to a second data communication network 214, to which different monitoring modules 210 and different diagnosing modules 212 can be connected.

During execution of the digital observer model 202, the remote data processing server 200 receives over the first data communication network 228 a digital input signal 205 reflecting the control input signals 222 to the production line 223, and a digital output signal 208 reflecting at least one of the operational states 224 of the production system 220.

The digital input signal 205 and digital output signal 208 are input to the digital observer model 202, which contains a model of production line 223, for example based on linear and/or non-linear state equations of the elements of the production line 223, and a model of automation system 226 containing copies of the control functions performed by the automation system 226. The digital observer model 202 is processed for a pre-determined period of time, which can be just one step in time, and thereby all operational states of the model of production line 223 and of the model of automation system 226, also called model states 207, are determined and - starting from an initial condition - updated.

The modeled states 207 are output via output interface 203 to the second data

communication network 214, where they can be accessed by monitoring modules 210 and/or diagnosing modules 221. The monitoring modules 212 are software applications running on external computer devices belonging to example to operators of the industrial plant, which may react to any abnormalities visible in the behavior of the modeled states by initiating and/or applying suitable counteractions 216, such as tuning of parameters in the automation system, or performing fault clearance. The diagnosing modules 221 are software applications running on different external computer devices belonging for example to service personnel of a supplier or sub-contractor of the industrial plant. The diagnosing of the modeled states may for example reveal that certain parts of the production line 223 require replacing or maintenance. The corresponding service actions 218 are then initiated and/or performed.

Instead of providing the digital output signal 224 directly to the digital observer model 202, it can first be input to a comparator 204 which is contained in the remote data processing server 200. In the comparator 204, the digital output signal 224 is compared to a

corresponding modeled state 206, i.e. a real operational state is compared to its simulated counterpart. The result of this comparison is then input to the digital observer model 202 as difference signal or correction signal 208 in order to be used for adapting the digital observer model 202.

A further alternative solution, which is indicated by a dashed extension of remote data processing server 200, is the implementation of at least parts of the monitoring modules 210 and of the diagnosing modules 212 inside the remote data processing server 200.

Consequently, the second data communication network 214 becomes an internal bus. Accordingly, the computer devices used by the operators and/or service personnel can be equipped with a reduced amount of processing capacity, as they only need to provide network access to a graphical user interface of the monitoring and/or diagnosing modules, for example a Web page.