FIELD OF INVENTION

[001] The present invention relates to an Intelligent Electronic Device (IED) of a substation automation system. More specifically, the invention relates to recommending an IED for a power system protection, control and automation functions,

BACKGROUND

[002] Substation automation comprises the monitoring, control, protection, and / or metering of various primary' equipment in the substation. Tire primary equipment could be electrical generators, electrical motors, power transformers, transmission and distribution lines, circuit breakers, capacitor banks, etc.

[003] lire substation automation is performed typically using intelligent electronic devices (IEDS) that generally receive electric power system information from primary equipment, make decisions (e g., trip decision to isolate and protect an electrical equipment) based on the information, and provide monitoring, control, protection, and/or automation functions.

[004] Selecting protection system for an equipment or plant is tedious. A protection scheme is provided generally by a customer where the protection scheme is analysed one or more IEDs are recommended. Generally, it is difficult to arrive at a single solution as there are vanous IEDs providing protection of different type. In existing protection systems, standard configurations or predefined application packages exist for every protection scheme, but further, customization or specific use cases has to be considered while selecting the right protection device. Conventional method of limiting the user to standard configuration and predefined application package lack flexibility from the user point of view and limits the user from selecting the optimum lED for the protection scheme,

[005] The protection system must adapt to a type of network making up the plant. Depending on the types of equipment and industrial process, protection functions to be selected can be different and sometimes not homogeneous with each other. Typically, before defining the protection systems, the network schemes are analysed, highlighting the advantages and disadvantages of the various solutions. [006] In the conventional systems, a list of protection system is provided as a static list depending on the type of protection and application of the protection. However, the static list may not be an optimized solution, as the conventional system do not consider the potential of protection system which can perform multiple protection functions.

[007] Hence, it is an objective of the present invention to provide an optimized solution of protection system that meets ail the requirement of the protection scheme and is compatible with the plant.

SUMMARY

[008] The above-mentioned shortcomings, disadvantages and problems are addressed herein winch will be understood by reading and understanding the following specification.

[009] The present invention provides a method for recommending an 1ED which provides protection to a substation according to a protection scheme provided by a customer device. IED configurations are generated on a computing system that could be in a cloud (a cloud computing system) or/and an edge system. The IED configurations generated on the edge or/and the cloud computing system are transmitted onto the physical IED enabling the physical IED to perform the application function corresponding to the IED configuration. Further, an IED whose configurations match with required configurations and having best performance metrics is identified and is recommended to the customer device.

[0010] The cloud computing system is provided with a virtual model of tire physical IED and is configured to be its digital twin. The generated IED configurations corresponds to one or more application functions of the physical IED. Tire application function lor the IED is typically identified or defined by a customer or an engineer. Examples of the IED application functions could involve performing a function of monitoring, control, metering, etc of a primary equipment of the substation or an entire bay of the substation.

[0011] The configurations corresponding to an IED application function are generated either in an automated mechanism or through a guided-engineering mechanism. The configurations corresponding to the IED application function are executed for each of a plurality of lElDs and are stored in a memory of the computing system. [0012] The computing system receives a request from a customer device for recommending a device for protecting one or more equipment in the substation. A customer provides a protections scheme via the customer device to the computing system. The computing unit determines at least one application function of the IED required for protecting the substation according to the protection scheme. The at least one application can be determined by referring to standards, interface with external networks, acceptable risk, maximum and minimum short circuit currents, status of neutral, presence of self-production in plant, coordination with existing system, configuration and network running criteria and practices. Further, die computing unit uses at least one machine learning techniques to search for application functions executed in existing IEDs and obtains configurations of lEDs having similar application functions. Furthermore, the computing unit identifies an IED from the plurality of IEDs having a best performance metrics. The identified IED is then recommended to the customer device.

[0013] The computing unit further recommends one or more application functions that can provide further protection to the substation based on the request. The computing unit uses the at least one machine learning technique to predict the one or more application functions that can provide further protection to the substation.

BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The appended drawings illustrate exemplary embodiments as disclosed herein and are not to be considered limiting in scope. In the drawings:

{0015} Figure 1 illustrates a physical device such as an IED and its functional representation, in accordance with some embodiments of the present invention;

[0016] Figure 2a illustrates functional representation of the physical device (such as IED) along with its alternate representation, in accordance with some embodiments of the present invention;

[0017] Figure 2b illustrates functional representation of the physical device implemented as a digital twin on the edge and cloud, in accordance with some embodiments of the present invention; [0018] Figure 3 illustrates complete system representation showing the critical components and the interaction between the various system components, in accordance with some embodiments of the present invention;

[0019] Figure 4 illustrates a representation of the primary elements involved in automated generation of the IED configurations as a sequence of functions and assigning of the input/output signals of sequence of functions;

[0020] Figure 5 illustrates a representation of the primary elements involved in generation of the IED configurations through a guided-engmeering mechanism of generating the sequence of functions and assigning of the inpu t/output signals of sequence of functions, in accordance with some embodiments of the present disclosure;

[0021] Figure 6 illustrates a summary of the method that represents recommending IED for protecting a substation according to a protection scheme, in accordance with some embodiments of the present disclosure.

DETAILED DESCRIPTION

[0022] Substation Automation (SA) systems include several basic SA functions for protection, control and monitoring of the substation. The automation functions can relate to either the individual constituents of primary equipment or to entire substation bays,

[0023] In power system, information is received by intelligent electronic devices (USDs) from the power apparatus (primary equipment) installed or/and from various sensors in the power system. The IEDS generate control commands that can maintain the system at normal operation.

[0024] In the following detailed description, reference is made to the accompanying drawings that form a part hereof, and which is shown by way of illustration specific embodiments, which may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the embodiments, and it is to be understood that other embodiments may be utilized. The following detailed description is, therefore, not to be taken in a limiting sense,

[0025] Figure la shows a representation of a typical physical device such as an IED and a mechanism to represent the physical device in a manner that facilitates constructing its virtual model. The virtual model is referred to herein as the digital twin since the virtual model is configured to replicate the functional operation of the physical device (the IED) under all test scenarios. [0026] Functionally, the physical device (IED) 100 can be represented as a system 1 10 that consists of a framework 135, application function engines 120, services 130, and communication Engine 125 as separate elements connected to the field over hardwired lines or over a bus. The framework 135 is defined as a collection of executable models that specify relevant functional aspects specific to the physical device (IED) 100. It formulates the structure and behaviour within a given application domain. The framework 135 also consists of a configuration manager 150, a data model unit 140, and a diagnostic unit 145.

[0027] The configuration manager 150 can facilitate in configuring various interactions between the application modules that constitute the functional representations of the physical device 100. Tire configuration manager 150 could also help with such processes as seting the precedence order of function module execution, synchronous and asynchronous aspects of parameters and data passing between various dependent functions.

[0028] A data model unit 140 organizes different data elements of the device (IED) and standardizes how the data elements relate with one another, lire data can be structured or unstructured. It can also be of any format, such as arbitrary' binary data, text, JSON, or serialized protocol buffers. The data elements could have components such as internet protocol (IP) address and MAC address of the physical device (lEDs), specific technical configuration information of the device (such as interrupt request lines, input/ output ports, etc), tag and identifier information associated with the device, hardware, software, and firmware versions, device model and state information (such as health of the device, current operational state

[0029] A diagnostic module 145 is a software component that facilitates monitoring various aspects of a device to provide system diagnostics. It can monitor parameters such as voltage, current, device internal temperature, etc.

[0030] The application function engine 120 is a component that facilitates build and run of application functions on the physical device or a computational unit embedded in the physical device. The various application fimctions enable performing device specific tasks such as a measurement of some parameter or control of an actuator etc. The communications engine 125 offers a platform for various connectivity options for data transfer among devices, between sensors and devices, between device and a gateway, between device and data management systems, etc. It enables using different network types such a wide area network (WAN), local area network (LAN), near-field communication (NFC) etc. and different communication protocols and standards such as Message Queuing Telemetry Transport (MQTT) etc.

[0031] The sendee modules 130 in the physical device can correspond to any of the sendee components such as error and failure diagnosis, health and performance monitoring, etc. The service module can integrate all the different sendee components to accomplish required sendee functions.

[0032] The abo ve described functional representation of the physical device (IED) 100 forms a complete description of the physical device 100 in terms of its functional components. However, in order to implement the digital twin of the physical device either on an edge device or/and on a cloud computing system, the functional representation is often further simplified to alternate representations. Such alternate representation enable further consolidation of various functionalities within the functional representation, based on specific aspects of the physical device whose functionality needs to be implemented and studied in the digital twin.

[0033] One such alternate representation is depicted in Figure 2. The physical model representation could be simplified either by consolidation or elimination of modules. Such simplified representation could be optimized by detecting and removing redundant attributes or modules. This would completely describe a physical device for certain applications or the physical devices that do not require detailed representations.

[0034] Such a representation provides the devices as consisting of a device model 235; a cognitive layer 230; the services module 220; and a training manager 240.

[0035] A device model 235 consists of the logical device that represents a functional view of the physical device. Functionally, it can be composed of units from framework that contains the data model unit, the diagnostic module, and the configuration manager,

[0036] The cognitive layer 230 consists of components to make application functions adaptive, using for instance machine learning algorithms to improve them based on the data they process in their specific environments. The cognitive layer 230 also facilitates programming intelligent applications and modules that operate in real environments, and in virtual environments that are designed to simulate real environments,

[0037] Above the cognitive layer 230, is representation of the application function engine, the communication engine, and few select services 225. [0038] Services module 220 forms the topmost layer of this logical representation. The service modules in the physical device can correspond to any of the service components such as error and failure diagnosis, health and performance monitoring, etc. The service module can integrate all the different service comp onents to accomplish required service functions

[0039] Training manager listens to (pre-processed) inputs and outputs and labels the samples for training leading to preparing probability tables or training result set. Labelling is performed on the training manager at the edge/cloud.

[0040] As shown in Figure 2b, the digital twin of the physical device (IED) 100 can be created on an edge device 210a or on a cloud computing system 210b. The communication between the physical device 100 and the digital twin on the edge device 210a can be accomplished through different communication protocols such as the Modbus TCP, IEC 61850 MMS (Manufacturing Message Specification), etc. Communication between the physical device 100 and the digital twin on die cloud computing system can be accomplished, for instance, via HTTP interface. Similarly, a communication between the digital twin on the edge device 210a and the digital twin on the cloud computing system 210b can be accomplished via HTTP interface.

[0041] The Configuration Data module 2,45 is provided to facilitate the IED digital twin to read and write all configuration and setting data of the corresponding physical IED, It also enables viewing and setting IED parameters

[0042] Figure 3 shows an embodiment of a representation of invention principles. The virtual model 215 of one more IEDS 320a, 320b are implemented on a cloud computing system 210b. Each IED has a corresponding digital twin 215 on the cloud computing system 210b. The IEDs 320a, 320b are configured to be in a bi-directional communication with their corresponding digital twins 215 as implemented on the cloud computing system 210b. The communication could be, for instance, via HTTP.

[0043] Each FED 320a, 320b could be part of a substation automation system where it provides an application function corresponding to an automation function comprising one or more of a protection, control and monitoring of the substation. The application functions of the IEDs 320a, 32Qn can correspond to automation of individual primary substation equipment 330a, 330n or to larger substation bays. The IEDs could either be existing IEDs 320a that are already commissioned and currently operational or they could also be new IEDs 320b that have been ordered or procured and need to be configured before getting commissioned for being operational for the substation automation

[0044] In order to configure the IEDs 320a, 320b for specific application functions or to add application functions to the lEDs, it is first required to obtain or specifically recognise one or more application functions that needs to be configured on any given or identified IED, The application functions can include automation functions such as protection, control and/or monitoring of the electrical equipment in a substation These automation functions can be enabled through one or more processing on the analog or/and digital signals received by the IED. The one or more processing could constitute harmonic detection, scheduling functions, specific waveform recording, etc. Such application functions enable generating configurations for the lEDs that facilitate adapting the IEDs for specific automation operation of electrical equipment in the substation.

[0045] The method for configuring the IEDs typically start with the cloud computing system 210b receiving a request for generating at least one configuration for the at least one application function in the IED from a customer device 300, wherein the request comprises an identifier associated with the IED 320a, 320b,

[0046] The customer device 300 could be a workstation such as an operator workstation or an engineering workstation within the substation. The customer device 300 could also be a portable handheld device such as a smartphone that can communicatively be connected with the cloud computing system 210b.

[0047] The identifiers associated with the IEDs 320a, 320b could consist of the IED serial number, IED’s IP address, lED’s media access control (MAC) address, firmware and software version in the IED, model number, manufacturer name, etc.

[0048] From the identifiers associated with the IEDs 320a, 320b, the cloud computing system determines the IED as at least one of an existing IED 320a in the substation and a new IED 320b procured for the substation. This determination is done by comparing the identifier associated with the IED that needs to be configured with identifiers of IEDs stored in the computing system 215 for generating the at least one configuration. The identifiers and configurations of IEDs stored in the computing system 215 correspond to every existing IED commissioned in the substation automation. If the comparison does not reveal any identical identifiers from the stored identifiers, the cloud computing system accepts the identified IED as a new IED 320b that has been procured for commissioning within the substation. Similarly, if the comparison does reveal any identical identifiers from tire stored identifiers, the cloud computing system accepts the identified IED as an existing IED 320a that is already commissioned within the substation and therefore the current configuration of this identified IED is correspondingly changed.

[0049] lEDs can he provided with customized or pre-configured application functions (such as protection, monitoring, control, metering, etc.) solutions for any type of primary substation equipment such as transformer, circuit breaker, etc. or/and for a substation bay.

[0050] Having identified the IED that need to be configured, the cloud computing system 215 adapts a virtual model 320at, 320bt for the IED by at least one of configuring a virtual model for the new IED as a digital twin to replicate functionalities of the new IED in the cloud computing system and using a pre-configured virtual model for the existing IED comprised m the computing system.

[0051] If the identified IED is one of the existing lEDs, the cloud computing system just adapts the virtual model of the existing IED that already exists and stored within the cloud computing system .

[0052] If the identified IED is one of the new lEDs, the cloud computing system first builds the virtual model of the new IED and stores this virtual model of the IED within the cloud computing system. The virtual model is built using standard programs and software environments such as digital simulators, relay modelling software, IED configurator software, etc. Tire cloud computing system can receive a vendor supplied IED capability description (ICD) file from the customer device 300. The ICD file provides the information needed by the cloud computing system to build the virtual model of the new IED. The information could consist of the number of the analog and digital channels, processing and storage capabilities, time synchronization methods such as internal clocks or/and GPS satellite clocks, etc.

[0053] Once the virtual model is built (for new lEDs 320b) or/and obtained (for existing IEDs 320a), the cloud computing system can now configure the virtual model for the IED as a digital twin to replicate functionalities of the new IED in the cloud computing system. [0054] Further details on generating the IED configurations for specifically requested application functions is presented with reference to Figures 4 and 5.

[0055] Figure 4 shows alternate representation of physical device on the cloud computing system 210b. The invention provides with a software configuration tool 400 that is hosted typically on tire customer device (300, Figure 3). The software configuration tool 400 receives, from a user/customer, a request for one or more application functions to be implemented on the IEDS. The application functions could be, for instance, monitoring, control , or/and metering of specific primary equipment in the substation. The application functions could also include other functionalities such as measurement of harmonies, power factor, etc. The application functions can correspond either to new IED configurations or to additional functionalities added to a given configuration.

[0056] The software configuration tool 400 can generate the IED configurations through, for instance, using a graphical language method by sequentially organizing various function blocks (graphical elements) that together constitute the configuration. The configuration file can be uploaded to the cloud computing system 210b in various file formats such as the JSON file format 410. Tire function blocks correspond to specific functions such as functions involved in analog to digital conversion processes, pre-processing on the analog or digital data, processing on the data, etc. Pre-processing on the analog or digital data could involve functions for filtering (high pass, low pass, band pass, etc.) the data, improving the signal-to-noise ratio, obtaining any estimated parameters from measured data values using one more algorithms, etc. Processing on the data could involve performing various functions such as fast Fourier transform, discreet Fourier transforms, wavelet transforms, data dimensionality reduction, functions for generating control signals, etc.

[0057] The above mentioned graphical language method could use a sequence prediction technique, which could be a machine learning based model, to facilitate generating of the IED configurations. The sequence prediction technique can perform simulations that include performing a similarity search to check for existing configurations that are similar to the required IED configurations. Such similarity search enables the sequence prediction technique to learn about the sequence of functions that need to be arranged in specific order. In addition to determining the order of arrangement for the sequence of functions, the sequence prediction technique also facilitates connecting the signals that correspond to input and outputs of the functions. The sequence prediction techniques sequentially determine the connections between at least one input and at least one output of the one or more functions. This too is performed using the similarity search performed above where the sequence prediction techniques can leam the input - output signal connections from historically used/stored configurations.

[0058] The cloud computing system 210b is configured to accept the JSON file 410 from the software configuration tool 400 hosted on the customer device (300, Figure 3). On the cloud computing system 210b, the Services module 220 contains the configuration function that processes tins JSON file. The configuration function within the sendees module 220 generates the binary files of the corresponding JSON files and writes into the framework 135 via ftp.

[0059] An application function block library (AFL) execution engine then executes this binary file which is the IED configuration corresponding to the FED application function. Any errors, warnings, or status of the binary execution by the AFL execution engine is then communicated 420 back to the software configuration tool 400 hosted on the customer device (300, Figure 3).

[0060] An alternate embodiment of the invention is represented in Figure 5 where a virtual agent or virtual assistant (also referred to herein as a chatbot) 510 is provided as an interface between an engineer/customer 500 and the cloud computing system 210b that hosts the digital twin of the IED and enables generating the configurations that is uploaded to the IED to configure them for specific application functions.

[0061] The IED configurations can be generated in an automated method or they could be generated through guided -engineering. Both the automated configuration generation and the guided-engineering of the IED configurations is enabled through the cognitive layer 230 of the IED functional representation.

[0062] In continued reference to Figure 5, an engineer/customer 500 starts by initiating a session with the virtual agent 510 where the virtual agent 510 could be hosted on any customer device (300, Figure 3) and is configured to interact with the engineer/customer 500 through text and speech interface.

[0063] An engineer/eustomer can issue a voice command or a text command to the virtual agent 510. The command could be in the natural language of the engineer/customer 500. Nature of the command could involve sending information to the virtual agent 510 on specific application functions for an IED configuration.

[0064] The virtual agent 510 then converts any unstructured commands such as that obtained from natural language of engineer/customer to a structured format using standard algorithm implementations of natural language process and speech-to-text etc. The virtual agent 510 then sends this structured format command to the cloud computing system 210b. The cloud computing system 210b uses the cognitive layer module 230 of the IED digital twin to engineer the IED configurations corresponding to the specified application function for the IED, The engineering of the IED configurations can be performed in one of two ways - through an automated mechanism or through a guided-engineering mechanism.

[0065] In the guided engineering mechanism of generating the IED configurations, various tools that are part of the cogniti ve layer module 230c of the implemented digital twin of the IED, facilitate in building the configurations in a step-by-step manner using graphical elements referred to as function blocks. Aspects of the function block were described earlier in discussion with reference to Figure 3.

[0066] Figure 5 further shows few relevant functions 230a, 230b, 230c performed by few tools within the cognitive layer module 230. Hie tools within the cognitive layer module 230 could constitute of various artificial intelligence and machine learning based algorithms/models typically employed for time-series forecasting. Examples of such tools could be gradient boosting models, recurrent neural networks (RNNs) models, long short-term memory (LSTM) networks, and the like that can enable performing a time-series forecasting or sequential predictions based on the input/output data of previous times or sequences.

[0067] In Figure 5, sequence forecasting ML algorithm on the cloud computing system 210b receives a command from the virtual agent 510.

[0068] The communication between the virtual agent 510 and the cloud computing system 210b could be through various web communication protocols such as the HTTP, HTTPS, etc.

[0069] The command could be to generate the IED configuration for enabling specific application functions such as monitoring, control, etc of specific primary equipment or a substation bay. Once the command is received by the cloud computing system 210b, the cognitive layer module 230 on the cloud computing system 210b processes this command. The processing involves searching the database of previously saved or archived IED configurations that matches in similarity to the IED configuration for the received application function. This is represented by 230a in Figure 5. Based on the similar configurations that the sequential forecasting machine learning algorithm finds, the sequential forecasting machine learning algorithm generates or recommends next. These are referred to as the machine learned configurations 230b. The machine learned configurations are then pushed to the virtual assistant 510 hosted on the customer device (300, Figure 3) as recommended next steps 230c. Thus, sequence of functions are built through the process of recommended next steps till the objective of the application function is achieved. This sequence of functions then constitutes the IED configuration corresponding to the requested application function.

[0070] The ML algorithm further recommends the IED to the customer which can protect the substation according to the protection scheme provided as input by the customer via the customer device. The ML algorithm determines the required application functions to protect the substation according to the protections scheme. Further, the ML algorithm considers the configurations of each existing plurality of !EDs and searches among the configured plurality of IEDs having the application functions which can protect eh substation according to the protection scheme. Thereafter, the ML model identifies an IED from the plurality of !EDs which has the highest performance metrics and recommends the IED to the customer device.

[0071] Further details on the specifics of the functions in the above described sequence of functions and their input/output configurations are provided in the discussion on the example application.

[0072] While the guided-engineering as described above is one mechanism for generating the IED configurations, the other mechanism is for automated generation of tire IED configurations. In the automated generation of the IED configurations, the sequential forecasting machine learning algorithm first receives, in a structured format, the command for specific application function. The algorithm then searches for configuration similar to the requested application function. This is represented by 230a in Figure 5.

[0073] In order to setup the digital twin of the IED on the cloud computing system, an estimate is first obtained on the cloud resources (processing and memory) needed to setup the digital twin and engineer the IED configurations. The estimation is performed by the software configuration tool hosted on the device interface. The estimation is obtained on the basis of the IED application functions that need to be configured and other configuration data of the IED. Various standard algorithms including supervised machine learning algorithms could be used for such resource estimation . The twin resources are estimated to consider the size of the device itself, additionally training manager and storage tor training manager, the storage for more training data, the considerations of sendees to be made available on the twin instance compared to the device. For a feeder device this could roughly be 3.5 times device size considerations and 2 times processing power. [0074] An edge or a cloud computing system is then prepared using die resources of memory and processing power as estimated in step 610. Using available services (like RESTful on Azure) a twin instance (the virtual model) is created by using Web Interfaces directly from web based/other engineering tools such as the software configuration tool,

[0075] The instance of the digital twin created is deployed on the edge or/and the cloud computing system. That is, deploy the device model of the physical device (such as the IED) and services including the configuration services.

[0076] Some configurations are pre-learned in the system based on previous usage experience of functions in other devices. The Configuration Training manager listens to Configuration Manager and based on the new configuration saved, can immediately initiate next step

[0077] The cloud computing system starts search for next steps based on Machine Learnt trained repository _'· - for instance, a Long Short Term Memory _' (LSTM) based Recurrent Neural Network that predict the next function block and connections between the function blocks. The Machine Learnt trained repository consists of one or sequence prediction techniques that can be used to further rank any generated predictions considering customer profile using Collaborative Filtering Algorithm.

[0078] The sequence prediction technique provides the next steps to configure the IED. The prediction of sequence of functions can be based on functions as a network approach. The recommendations for the sequence of functions arc based on user acceptance goes to next steps in engineering and keeps informing the system acceptance like epu (processor) load estimation to user over the interface.

[0079] Referring to Figure 6, at step 601, the computing unit receives the request for recommending the IED for protecting the computing system The protection scheme comprises a specification of the substation, details of the one or more electrical equipment 330a, 330b, a protection type for each electrical equipment and a protection zone. The specification of the substation can be in a format such as a single line drawing (SLD) or any other format which provides information about the substation. The details of the one or more electrical equipment 330a, 330b can include datasheet of the electrical equipment. The protection type can include overcurrent protection, overvoltage protection, undervoltage protection, flame protection, fire protection and the like. The protection zone includes zone protection, machine protection, selective protection, non-selective protection, protection in support, interface protection, and the like. [0080] At step 602, the computing system determines at least one application function of the IED required to protect the substation according to the protection scheme. The determination can be made based on the protection scheme provided by the customer. The at least one application can be determined by referring to standards, interface with external networks, acceptable risk, maximum and minimum short circuit currents, status of neutral, presence of self-production in plant, coordination with existing system, configuration and network miming criteria and practices. In an embodiment, the acceptable risk can be a measure provided by the customer about the consequence of a fault in the power system and how sever the power system is. Likewise, the customer may provide the maximum and minimum short circuit current values, outside which the power system fails. The network running criteria may refer to type of network used in the customer premises. For example, the customer premise may have a single radial network. However, the selected IED may be compatible with a double radial network. Hence, such IEDs cannot be recommended, and one or more lEDs that are compatible with the single radial network are recommended. The recommendation of the IEDs in tire above examples are by considering only the parameters mentioned in the examples. A person skilled will appreciate that a final IED recommended to the customer will consider oilier parameters and an overall performance metrics. The standards can be used to determine the at least one application function. For example, for providing overcurrent protection for a generator, an application of an IED which can monitor current value of a line and can trip a circuit when the current value is more than an expected value is selected. Likewise, for all the protection defined in the protection scheme, the application functions are determined.

[0081] At step 603 the virtual model is adapted for generating configurations for each existing IED. Generating the configurations for the IED is explained in detail in the above paragraphs.

[0082] At step 604 the computing system obtains the configurations of the plurality of IEDs on which the determined application functions are executed. In an embodiment, the configurations of IEDs which support the determined application functions are only obtained.

[0083] At step 604, the ML algorithm identifies the IED among the plurality of IEDs for providing protection to the substation according to the protection scheme. The ML algorithm identifies the IED based on the configurations of the plurality of IEDs, and performance metrics of the plurality of IEDs. The performance metrics includes but are not limited to, configuration time, CPU usage, localization, compatibility and parameter setting. In an embodiment, the IED having the highest performance metrics is identified. For example, for a protection scheme which requires overcurrent protection of a generator, undervoltage protection and temperature protection, the IED having configurations which can perform all the above applications are identified. Further, based on the configurations, the performance metrics are evaluated. Still referring to the example, when two IEDs are identified, the IED having the best performance is selected tor recommendation. A first IED may have less CPU usage compared to the second IED. In an embodiment, other parameters such as network compatibility, compatibility with other equipment in the substation, localization, are also considered for identifying the IED to be recommended. In an embodiment, a plurality of IEDs may be recommended based on the performance metrics. In an embodiment, when a single IED is not able to provide the required configurations with high performance, other IEDS may be recommended such that the performance of providing the required configuration is high. For example, consider fifty functions are required to be executed by a selected IED. However, considering CPU execution time of each function, executing the fifty application may overload the selected IED. Hence, the computing system may recommend two or more IEDs such that the fifty applications may be distributed among the two or more IED to meet the performance criteria. Likewise, other performance metrics may also be considered to recommend the two or more IEDs. In an embodiment, the computing unit may further recommend a sequence of executing the fifty applications by the two or more IEDs.

[0084] The ML algorithm is trained to create dataset and configurations that provide protection to the substation. During the training, the ML algorithm creates dataset comprising the required application functions for different sets of inputs. For example, for different protection scheme, different application functions are determined. Tire ML algorithm can undergo feedback techniques to improve the efficiency of the dataset. Likewise, the ML algorithm can create configurations dataset based on the application function dataset using the configuration data of the plurality of IEDs. Further, the MI, algorithm classifies the dataset and the configurations based on a weight factor and performance metrics. For example, the overcurrent protection can have a higher weight compared to thermal protection. Hence, the IEDs providing overcurrent protection are classified together. Further, when IEDs are configured with both the overcurrent application and the temperature application, the IEDs having the primary function with the highest weight are classified together. Thereafter, the ML algorithm extracts features from the classified set. The features can be performance metrics having highest values. The extracted features are used to identify the IED for recommending to the customer. The ML algorithm can also select a plurality of IEDs when a single IED cannot he identified from the cluster providing all the required configurations.

[0085] At step 605, the ML, algorithm recommends the identified IED (or a plurality of lEDs) to the customer device. The recommendation may he provided via the user interface (chat box) which may be at the customer device.

[0086] In an embodiment, the ML algorithm can include clustering techniques. Naive Bayes technique. The ML algorithm can predict one or more application functions for providing further protection to the substation based on the request using the at least machine leaming technique and recommend to the customer device, the one or more application functions for configuring in the recommended IED.

[0087] The present invention recommends optimized IED (or plurality of lEDs) to protect the substation. Tire ML algorithm identified the best IED which can provide multiple protection, unlike conventional systems which provide static list of lEDs. Further, as the digital twin is used, different configurations are generated by executing different application functions. Hence, the optimized IED can be identified which meets the requirements of tire protection scheme.

[0088] This writen description uses examples to describe the subject matter herein, including the best mode, and to enable any person skilled in the art to make and use the subject matter. The patentable scope of the subject matter is defined by the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with insubstantial differences from the literal language of the claims.