A system and a method for detecting a type of power source

Abstract

A PSIS (100) for detecting a type of power source connected to an electric load (108) via a load bus (102), is disclosed. The PSIS 100 includes a sensor unit (208), a filtering unit (210), and a detection unit (214). The filtering unit is adapted to process each set of a current signals and a voltage signals generated by the sensor unit (208) to filter noise therein and generate a plurality of output values corresponding to a filtered set of the current signals and the voltage signals. The detection unit (214) is adapted to process the plurality of output values to identify a pattern using a supervised machine learning model. The detection unit (214) is adapted to compare the identified pattern with a set of previously learned patterns by the supervised machine learning model corresponding to different types of power sources (106) to identify the type of power source.

Description

[0001] A SYSTEM AND A METHOD FOR DETECTING A TYPE OF POWER SOURCE

[0002] FIELD OF THE INVENTION

[0003] The present disclosure relates to a system and more particularly, relates to a power source identification system and a method for detecting a type of power source connected to an electric load via a load bus.

[0004] BACKGROUND

[0005] A plurality of power sources is adapted to provide a DC output power, to power an electric load. Typically, the plurality of power sources can include, but is not limited to, a battery, a power grid, etc. Generally, most of the power sources rely on converting alternating current (AC) power to DC power with the help of a rectifier. The DC output power from a power source is then supplied to the electric load through a load bus. A unidirectional flow of the DC output power is reliable for powering the electric load.

[0006] However, the DC output power generated from the plurality of power sources has various harmonics, i.e., minor fluctuations in the DC voltage, which can hamper the performance of the electrical load. In some cases, the harmonics may gradually decrease the life and performance of the electric load. Further, the harmonics as introduced in the DC output power also damages the electric load. Accordingly, there is a need to identify an input power source at a point / j unction of the DC output power. The identification of the input power source is important as now electronics have integrated with electrical networks. Further, detection of the input power source may be done at a power management system or a power switching unit which supplies power to the electric load. However, such configuration requires dedicated equipment and is non-compatible with the already existing system.

[0007] Hence, there is an immense need to provide a system and a method for detecting a type of power source at a junction of a DC output power while overcoming abovementioned problems.

[0008] SUMMARY

[0009] This summary is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the invention. This summary is neither intended to identify key or essential inventive concepts of the invention and nor is it intended for determining the scope of the invention.

[0010] The present disclosure aims to provide a system and a method for detecting a type of power source from different types of power sources using an artificial intelligence / a machine learning approach.

[0011] In an embodiment of the present disclosure, a power source identification system (PSIS) for detecting a type of power source connected to an electric load via a load bus is disclosed. The PSIS includes a sensor unit, a filtering unit, and a detection unit. The sensor unit is connected to the load bus. The sensor unit is adapted to measure amperage and voltage of the electric current flowing through the load bus to generate a set of current signals and voltage signals within a predetermined time, respectively. The filtering unit is adapted to process each set of the current signals and the voltage signals to filter noise therein and generate a plurality of output values corresponding to a filtered set of the current signals and the voltage signals. The detection unit is adapted to process the plurality of output values to identify a pattern using a supervised machine learning model. The detection unit is adapted to compare the identified pattern with a set of previously learned patterns by the supervised machine learning model corresponding to different types of power sources to identify the type of power source.

[0012] In another embodiment, a method for detecting a type of power source connected to an electric source via a load bus through a power identification system (PSIS) is disclosed. The method includes sensing, by a sensor unit, the electric current and measuring amperage and voltage of the electric current flowing through a load bus to generate a set of current signals and voltage signals within a predetermined time interval, respectively. The method includes filtering, by a filtering unit, noise from each set of the current signals and the voltage signals therein, and generating, by the filtering unit, a plurality of output values corresponding to a filtered set of current signals and voltage signals. The method includes processing, by a detection unit, the plurality of output values to identify a pattern using a supervised machine learning model. The method includes comparing, by the detection unit, the identified pattern with a set of predetermined patterns corresponding to different types of power sources to identify the type of power source.

[0013] The present disclosure provides a configuration of the PSIS, where the PSIS is adapted to identify the type of power source at a place close to the load bus instead of a power management system. This configuration of the PSIS ensures enhanced accuracy in identifying the type of power source. The present configuration also restricts accessing a power management system, i.e., provides leverage to the electric load, where the electric load may identify activity of any input power source, i.e., instants of switchover, source label, and all other details that can be generated as a byproduct, without having to tap into a system which could be outside jurisdiction of the electric load. The configuration as disclosed ensures efficient deduction of a DC load pattern and a harmonic source identification. The configuration ensures detection of switchover to different power sources based on the identified harmonics pattern by a power source in the electric load through the load bus. This configuration eliminates impact of the harmonics in the electric load through the load bus, by ensuring that necessary actions are implemented to improve harmonic profile of a given input power source, thus ensuring enhanced life and performance of the electric load.

[0014] To further clarify the advantages and features of the present invention, a more particular description of the invention will be rendered by reference to specific embodiments thereof, which is illustrated in the appended drawings. It is appreciated that these drawings depict only typical embodiments of the invention and are therefore not to be considered limiting of its scope. The invention will be described and explained with additional specificity and detail with the accompanying drawings.

[0015] BRIEF DESCRIPTION OF THE DRAWINGS

[0016] These and other features, aspects, and advantages of the present invention will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:

[0017] Figure 1 illustrates a block diagram of a PSIS along with a load bus, different types of power sources, and an electric load, according to an embodiment of the present disclosure;

[0018] Figure 2 illustrates a block diagram of the PSIS, according to an embodiment of the present disclosure;

[0019] Figure 3 illustrates a block diagram of the PSIS where a training unit and a detection unit of the PSIS is a cloud-based server unit, according to an embodiment of the present disclosure; Figure 4 illustrates a method performed by the PSIS, according to an embodiment of the present disclosure;

[0020] Figure 5 illustrates a training of a supervised machine learning model of the PSIS, according to an embodiment of the present disclosure;

[0021] Figure 6 illustrates an explanation of a data separation using a supervised machine learning model of the PSIS, according to an embodiment of the present disclosure;

[0022] Figure 7A illustrates a graph showing a precision versus a training length of the PSIS, according to an embodiment of the present disclosure;

[0023] Figure 7B illustrates a graph showing an accuracy versus a bandwidth of a kernel function of the PSIS, according to an embodiment of the present disclosure;

[0024] Figure 8A illustrates a graph showing an accuracy versus a training length at a predetermined sampling frequency of the PSIS, according to an embodiment of the present disclosure;

[0025] Figure 8B illustrates a graph showing a sensitivity versus a sampling frequency at the training length of the predetermined samples of the PSIS, according to an embodiment of the present disclosure;

[0026] Figure 8C illustrates a graph showing an accuracy versus a measurement noise at the training length of the predetermined samples, and the bandwidth of the kernel function of the PSIS, according to an embodiment of the present disclosure;

[0027] Figure 9A illustrates a graph showing a type of power source from different types of power sources and a switchover identification in voltage measurements of the PSIS, according to an embodiment of the present disclosure;

[0028] Figure 9B illustrates a graph showing a type of power source from different types of power sources and a switchover identification in current measurements of the PSIS, according to an embodiment of the present disclosure; Figure 10A illustrates a graph showing an accuracy versus a training length of the PSIS in a laboratory test set-up and a field test set-up, according to an embodiment of the present disclosure;

[0029] Figure 10B illustrates a graph showing an accuracy versus a sampling frequency of the PSIS in the laboratory test set-up and the field test set-up, according to an embodiment of the present disclosure; and

[0030] Figure IOC illustrates a graph identifying the type of power source from different types of power sources using the PSIS in the laboratory test set-up and the field test set-up, according to an embodiment of the present disclosure.

[0031] Further, skilled artisans will appreciate that elements in the drawings are illustrated for simplicity and may not have necessarily been drawn to scale. Furthermore, in terms of the construction of the device, a plurality of components of the device may have been represented in the drawings by conventional symbols, and the drawings may show only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the drawings with details that will be readily apparent to those of ordinary skill in the art having benefit of the description herein.

[0032] DETAILED DESCRIPTION OF FIGURES

[0033] For the purpose of promoting an understanding of the principles of the invention, reference will now be made to the embodiment illustrated in the drawings and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the invention is thereby intended, such alterations and further modifications in the illustrated system, and such further applications of the principles of the invention as illustrated therein being contemplated as would normally occur to one skilled in the art to which the invention relates. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skilled in the art to which invention belongs. The system and examples provided herein are illustrative only and not intended to be limiting.

[0034] For example, the term “some” as used herein may be understood as “none” or “one” or “more than one” or “all.” Therefore, the terms “none,” “one,” “more than one,” “more than one, but not all” or “all” would fall under the definition of “some.” It should be appreciated by a person skilled in the art that the terminology and structure employed herein is for describing, teaching, and illuminating some embodiments and their specific features and elements and therefore, should not be construed to limit, restrict or reduce the spirit and scope of the present disclosure in any way.

[0035] For example, any terms used herein such as, “includes,” “comprises,” “has,” “consists,” and similar grammatical variants do not specify an exact limitation or restriction, and certainly do not exclude the possible addition of a plurality of features or elements, unless otherwise stated. Further, such terms must not be taken to exclude the possible removal of the plurality of the listed features and elements, unless otherwise stated, for example, by using the limiting language including, but not limited to, “must comprise” or “needs to include.”

[0036] Whether or not a certain feature or element was limited to being used only once, it may still be referred to as “plurality of features” or “plurality of elements” or “at least one feature” or “at least one element.” Furthermore, the use of the terms “plurality of’ or “at least one” feature or element do not preclude there being none of that feature or element, unless otherwise specified by limiting language including, but not limited to, “there needs to be plurality of... ” or “plurality of elements is required.”

[0037] Unless otherwise defined, all terms, and especially any technical and / or scientific terms, used herein may be taken to have the same meaning as commonly understood by a person ordinarily skilled in the art.

[0038] Reference is made herein to some “embodiments.” It should be understood that an embodiment is an example of a possible implementation of any features and / or elements of the present disclosure. Some embodiments have been described for the purpose of explaining plurality of the potential ways in which the specific features and / or elements of the proposed disclosure fulfill the requirements of uniqueness, utility, and non-obviousness.

[0039] Use of the phrases and / or terms including, but not limited to, “a first embodiment,” “a further embodiment,” “an alternate embodiment,” “one embodiment,” “an embodiment,” “multiple embodiments,” “some embodiments,” “other embodiments,” “further embodiment”, “furthermore embodiment”, “additional embodiment” or other variants thereof do not necessarily refer to the same embodiments. Unless otherwise specified, plurality of particular features and / or elements described in connection with plurality of embodiments may be found in one embodiment, or may be found in more than one embodiment, or may be found in all embodiments, or may be found in no embodiments. Although plurality of features and / or elements may be described herein in the context of only a single embodiment, or in the context of more than one embodiment, or in the context of all embodiments, the features and / or elements may instead be provided separately or in any appropriate combination or not at all. Conversely, any features and / or elements described in the context of separate embodiments may alternatively be realized as existing together in the context of a single embodiment.

[0040] Any particular and all details set forth herein are used in the context of some embodiments and therefore should not necessarily be taken as limiting factors to the proposed disclosure.

[0041] Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

[0042] Figure 1 illustrates a block diagram of a PSIS 100 along with a load bus 102, different types of power sources 106, and an electric load 108, according to an embodiment of the present disclosure. In an embodiment, the different types of power sources 106 may be electrically communicated to a DC power management unit 104. In an embodiment, the different types of sources 106 may be source 1, source 2, and source 3, without departing from the scope of the present disclosure . In another embodiment, the different types of sources 106 may be N number of sources, without departing from the scope of the present disclosure. In an embodiment, the different types of power sources 106 may be an inverter, a battery, a GenStat, a transformer, and a power grid, without departing from the scope of the present disclosure. Further, the DC power management unit 104 may be electrically connected to the electric load 108 through the load bus 102. The different types of power sources 106 may be adapted to generate power which may be transmitted to the DC power management unit 104. The DC power management unit 104 may be adapted to transfer the power to the electric load 108 through the load bus 102.

[0043] In an embodiment, the electric load 108 may be a cellular tower, or a TV tower without departing from the scope of the present disclosure. In an embodiment, the load bus 102 may be a DC load bus, without departing from the scope of the present disclosure. Further, the PSIS 100 may be disposed at an end of the load bus 102. The PSIS 100 may be adapted to identify a type of power source from the different power sources which provide power to the electric load 108 through the DC power management unit 104 and the load bus 102. Further, the PSIS 100 may be a non-invasive system which uses a supervised machine learning model to identify the type of power source, without departing from the scope of the present disclosure. This configuration ensures an optimum and continuous DC output power from the type of power source amongst the different types of power sources 106 to the electric load 108.

[0044] Further, constructional features of the PSIS 100 may be explained in conjunction with Figures 2 and 3.

[0045] Figure 2 illustrates a block diagram of the PSIS 100, according to an embodiment of the present disclosure. Figure 3 illustrates a block diagram of the PSIS 100 where a training unit 216 and a detection unit 214 of the PSIS 100 may be a cloud-based server unit, according to an embodiment of the present disclosure.

[0046] The PSIS 100 may include, but is not limited to, a sensor unit 208, a filtering unit 210, and the detection unit 214. The PSIS 100 may also include a processor / controller 202, a memory 204, unit(s) 206, and data 218. The memory 204, in one example, may store the instructions to carry out the operations of the units 206. The units 206 and the memory 204 may be coupled to the processor 202.

[0047] The processor 202 may be a single processing unit or several units, all of which could include multiple computing units. The processor 202 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processor, central processing units, state machines, logic circuitries, and / or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor 202 is configured to fetch and execute computer-readable instructions and data stored in the memory 204.

[0048] The processor 202 may include one or a plurality of processors. At this time, one or a plurality of processors may be a general-purpose processor, such as a central processing unit (CPU), an application processor (AP), or the like, a graphics-only processing unit such as a graphics processing unit (GPU), a visual processing unit (VPU), and / or an Al-dedicated processor such as a neural processing unit (NPU). The one or a plurality of processors control the processing of the input data in accordance with a predefined operating rule or artificial intelligence (Al) model stored in the non-volatile memory and the volatile memory. The predefined operating rule or machine learning model is provided through training or learning.

[0049] The memory 204 may include any non-transitory computer-readable medium known in the art including, for example, volatile memory, such as static random-access memory (SRAM) and dynamic random-access memory (DRAM), and / or non-volatile memory, such as read-only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes.

[0050] The units 206, amongst other things, include routines, programs, objects, components, data structures, etc., which perform particular tasks or implement data types. The units 206 may also be implemented as, signal processor(s), state machine(s), logic circuitries, and / or any other device or component that manipulates signals based on operational instructions.

[0051] Further, the units 206 can be implemented in hardware, instructions executed by a processing unit, or by a combination thereof. The processing unit can comprise a computer, the processor 202, such as the processor 202, a state machine, a logic array, or any other suitable devices capable of processing instructions. The processing unit may be a general -purpose processor which executes instructions to cause the general-purpose processor to perform the required tasks or, the processing unit can be dedicated to performing the required functions. In another embodiment of the present disclosure, the units 206 may be machine-readable instructions (software) which, when may be executed by the processor 202, perform any of the described functionalities. Further, the data serves, amongst other things, as a repository for storing data processed, received, and generated by one or more of the units 206. The data 218 may include information and / or instructions to perform activities by the processor 202.

[0052] The unit(s) 206 may perform different functionalities in the PSIS 100. Accordingly, the unit(s) 206 may include the sensor unit 208, the filtering unit 210, a transformation unit 212, the detection unit 214, and the training unit 216.

[0053] In an embodiment, the sensor unit 208 may be connected to the load bus 102. Further, the sensor unit 208 may include a voltage sensor and a current sensing unit. The voltage sensor may be connected across an end of the load bus 102. Further, the current sensing unit may be adapted to connect to the load bus 102 in a non-invasive manner.

[0054] The current sensing unit may be a hall effect sensor, a shunt resistor, a hall effect current probe, current sensing integrated circuits, or a magnetic field sensor, without departing from the scope of the present disclosure. In another embodiment, a current transformer, Rogowski coils may be adapted to measure the amperage of the electric current, where the electric current may be AC current, without departing from the scope of the present disclosure. In one example, the filtering unit 210 may be connected to the sensor unit 208. The transformation unit 212 may be connected to the filtering unit 210. Further, the detection unit 214 and the training unit 216 are subsequently connected with the transformation unit 212. In an embodiment, the sensor unit 208, the filtering unit 210, the transformation unit 212, the detection unit 214, and the training unit 216 may be configured in the PSIS 100 in various configurations, without departing from the scope of the present disclosure.

[0055] Referring to Figure 2, the PSIS 100 may be a standalone system, where the sensor unit 208, the filtering unit 210, the detection unit 214, and the training unit 216 may be adapted to be disposed in proximity of the load bus 102, without departing from the scope of the present disclosure. Further, the transformation unit 212 and the training unit 216 may be also disposed in the proximity of the load bus 102, without departing from the scope of the present disclosure .

[0056] In another embodiment, referring to Figure 3, the sensor unit 208, the filtering unit 210, may be adapted to disposed in proximity of the load bus 102, without departing from the scope of the present disclosure. Further, the detection unit 214 may be the cloud-server unit. Further, the detection unit 214 may be adapted to be communicatively coupled with the sensor unit 208 and the filtering unit 210. In another embodiment, the transformation unit 212 may be disposed in proximity of the load bus 102, without departing from the scope of the present disclosure. Further, in another embodiment, the training unit 216 may be the cloud-server unit along with the detection unit 214, without departing from the scope of the present disclosure.

[0057] In an embodiment, the PSIS 100 may be adapted to identify the type of power source from the different types of power sources 106. Further, the PSIS 100 may be adapted to communicate the identified type of power source to a central database management system (CDMS) 110, where the CDMS 110 may be communicatively coupled with the PSIS 100 (as shown in Figure 1). Particularly, the detection unit 214 may be adapted to communicate identified patterns to the CDMS 110 through one of a wired or a wireless network, without departing from the scope of the present disclosure.

[0058] In an embodiment, the PSIS 100 stores data, that is an identified pattern adapted to identify the type of power source, in a predetermined format. In an embodiment, the predetermined format may be a master-text file, without departing from the scope of the present disclosure. Further, the data updates with every new recorded measurement. Three threads may be executed in parallel at the PSIS 100 to avoid delays pertaining to series execution of tasks.

[0059] The first thread realizes data capturing and storage in the master text file, the second thread undertakes source classification and data storage into multiple excel sheets, and the third thread achieved communication to a server, conventionally on a different network. A flexibility may be provided in choosing any refresh rate beyond 1 minute. This threshold on the refresh rate may be imposed to avoid overlap among the multiple threads being executed at the PSIS 100. For a refresh rate of R, the data for every R minute(s) may be stored in a new excel files for transmission. Finally, the data, as stored, may be pushed onto CDMS 110 through a communication architecture by following a communication protocol.

[0060] In an embodiment, the communication architecture may include a pair of components. For example, a client and the server. The PSIS 100 may support connectivity of a predetermined wifi band. The predetermined wifi bands may be over 2.4 GHz and 5 GHz Wi-Fi bands using IEEE 802. 11.b / g / n / ac wireless LAN standard. The PSIS 100 acts as the client, responsible for transmitting the excel files, while the CDMS 110 acts as the server, responsible for receiving the excel files. The communication between the client and the server may be established through the internet using TCP / IP network protocol. The CDMS 110 may be capable of serving multiple PSIS clients simultaneously using the multi-threading parallel-processing ability at the CDMS 110. A detailed implementation of the communication between PSIS 100 and the CDMS 110 and the multi -access at the CDMS 110 may be discussed in subsequent paragraphs.

[0061] In an embodiment, the communication protocol may be configured by establishing a wired and or wireless network on the PSIS 100. Further, a router may be configured on the PSIS-enabling port and forward a port number used for communication. Further, a server script may be configured on the CDMS 110. The configuration may include a socket object which may be created and further, binding the socket object to an IP address and the port number. The server script may be adapted to respond to incoming connections and receive the transmitted data. Further, a client script may be configured on the PSIS 100. The configuration involves creation of a socket object. Further, the socket object may be connected to an IP address of the server and the port number. Further, the socket object transmits the excel files using sendal method. Further, the communication may be examined by running both the client and the server scripts. Further, the verification of transmitting and receiving the excel files may be completed. In an embodiment, as the PSIS 100 and CDMS 110 may be on different networks, port forwarding may be configured on the router of CDMS 110 to enable incoming connections from the internet. This configuration allows the server script on the CDMS 110 to receive the transmitted data from the client on the PSIS 100. The excel files may be transmitted as a stream of bytes over the network connection. The server script receives the data stream and deduces the data stream to a file.

[0062] In an embodiment, the CDMS 110 supports multiple clients simultaneously. For instance, the CDMS 110 creates a TCP server. The server creates a main thread which creates a socket and binds the socket to a specified IP address and port number. Further, the thread function enters into the infinite loop, waiting for incoming connections. A scheduled PSIS 100 compresses the data using a zone improvement plan protocol and transmits to the CDMS 110. When a new connection may be accepted, a new sub-thread is created to handle the communication with the client. Further, the main thread goes back to waiting for additional connections. The sub-thread receives the data from the client and construct a unique filename for the file being received. A ‘writer script’ at the CDMS 110 decompresses a ZIP -compressed data and writes the received data to a newly created file until the connection is closed or no more data is received. This duration can be user-defined or set to default. If the ‘connection close’ duration is met, the CDMS 110 prints a message indicating that the file has been received from the client and then closes the connection. Further, multiple such operations may be run parallel in various threads, thus enabling the parallel processing, resulting in handling of the multiple clients simultaneously.

[0063] Further, the received data may be stored in the excel files on the CDMS 110 with the files being named in a predetermined format, for example, “SiteX DDMMYYYY HHMM TT”. In an embodiment, “SiteX” describes that the data being sent may be from the PSIS 100 installed at site X. In an embodiment, “DDMMYYYY” defines the date of the recorded data in DD / MM / YYYY format. In an embodiment. “HHMM” defines a start time for the data in the excel file may be in 24-hour format. Further, in an embodiment “TT” describes the duration of data recorded in that dataset, i.e., the refresh rate set by the user.

[0064] The excel sheets may include a time-stamped data of the voltage, the current, and the identified type of power source from the different types of power sources 106. The key utilization of this data logging is to facilitate a central data management for different distributed sites of the electric load 108. This source-tagged data can be used to understand billing, load distribution, harmonic source identification, etc. Further, the identification of the type of source and switchover between the different type of power sources in the present subject matter may be explained with reference to an example for providing more clarity.

[0065] For instance, default supply of the power to the load bus 102 may be maintained through the source 1, which in-tum also charges the source 2 to a pre-defined voltage Vo. Further, the identification of the type of source from the different types of sources 106providing power to the electric load 108 may be continuously determined by the PSIS 100 along with the CDMS 110 by executing the working which will be explained in subsequent paragraphs. Thus, in the event of failure or in any other instance, the type of the power source may switch to another type of power source from the different types of power sources 106 to supply the power immediately and continuously to the electric load 108. For an instance, in the event of failure of the source 1 or the detection of harmonics from the source 1, the PSIS 100 may be adapted to detect switchover such that the power may be supplied from the source 2 through a power management block (not shown), maintaining an uninterrupted power supply. This also acts as a backup source during source switchover. The source 2 remains functional until an r / fraction depth of discharge is attained, leading to a cutoff voltage Vc= r)V0. Once this condition is met, the source may switch to ‘source 3 ’, which powers the electric load 108, through the DC power management unit 104.

[0066] Further, working of the PSIS 100 for identifying the type of power source connected to the electric load 108 via the load bus 102, may be disclosed by a method 400.

[0067] Figure 4 illustrates the method 400 for detecting the type of power source connected to the electric load 108 via the load bus 102 through the PSIS 100. Figure 5 illustrates a training of a supervised machine learning model of the PSIS 100, according to an embodiment of the present disclosure. Figure 6 illustrates an explanation of a data separation using a supervised machine learning model of the PSIS 100, according to an embodiment of the present disclosure. The order in which the method steps are described below is not intended to be construed as a limitation, and any number of the described method steps can be combined in any appropriate order to execute the method or an alternative method. Additionally, individual steps may be deleted from the method without departing from the spirit and scope of the subject matter described herein.

[0068] The method 400 can be performed by programmed computing devices, for example, based on instructions retrieved from non-transitory computer-readable media. The computer-readable media can include machine -executable or computer-executable instructions to perform all or portions of the described method. The computer-readable media may be, for example, digital memories, magnetic storage media, such as magnetic disks and magnetic tapes, hard drives, or optically readable data storage media.

[0069] The method 400 is explained in conjunction with Figures 4, 5, and 6. The method 400 begins at block 402, includes sensing, by the sensor unit 208, an electric current and measuring the amperage and the voltage of the electric current flowing through the load bus 102 to generate a set of current signals and voltage signals within a predetermined time interval, respectively. The sensor unit 208 may be adapted to measure the amperage and the voltage of the electric current through the load bus 102 to generate the set of current signals and voltage signals with the predetermined time interval, respectively.

[0070] In one example, the sensor unit 208 samples the amperage and the voltage of the electric current, flowing through the load bus 102 at a predetermined sampling rate of up to 1000 samples per second to generate the set of current signals and voltage signals within the predetermined time interval. Particularly, the voltage sensor of the sensor unit 208 may be adapted to measure the voltage of the electric current. Similarly, the current sensing unit may be adapted to measure the electric current, that is, the amperage of the electric current flowing through the load bus 102.

[0071] Further, a voltage divider may be adapted to step-down values of the voltage as measured by the voltage sensor. In an embodiment, the electric current generates a magnetic field through which the amperage of the electric current may be measured by the current sensing unit ensuring a non-invasive and retrofittable PSIS 100. This configuration eliminates a requirement to disconfigure the electric load 108. Further, an analog -to-digital converter quantizes the reduced value of the voltage and the measured amperage of the electric current and provides the amperage and the voltage for the sampling to generate the set of current signals and voltage signals at the variable predetermined sampling rate. Further, the set of current signals and voltage signals may be time-stamped and provided to the filtering unit 210.

[0072] At block 404, the method includes filtering, by the filtering unit 210, noise from each set of the current signals and the voltage signals therein and generating, by the filtering unit 210, a plurality of output values corresponding to a filtered set of current signals and voltage signals. The filtering unit 210 may be adapted to process each set of the current signals and the voltage signals to filter noise therein. In an embodiment, the filtering unit 210 may be a moving average filtering unit, without departing from the scope of the preset disclosure. The filtering unit 210 may be adapted to receive the timestamped set of current signals and voltage signals. The filtering unit 210 may be adapted to filter the noise from the timestamped set of current signals and voltage signals. The filtering unit 210 may also generate a plurality of output values corresponding to the filtered set of the current signals and the voltage signals.

[0073] For instance, the filtering unit 210 may average each set of the current signals and the voltage signals to generate a transient impedance corresponding to each set of the current signals and the voltage signals. The filtering unit 210 may be adapted to average each set of the current signals and the voltage signals over a sliding window of a fixed length that moves over a data stream in a time-division multiplexed manner and further generates the transient impedance. In an embodiment, the voltage signals may be in a range of 52V to 48V, without departing from the scope of the present disclosure. In one embodiment, the current signals may be 1 ampere, without departing from the scope of the present disclosure. Further, the transient impedance as generated may be in a range of 52 to 48 . without departing from the scope of the present disclosure. Further, the filtering unit 210 may be adapted to filter the noise from the transient impedance.

[0074] Further, a weight may be allocated to the transient impedance, such that the transient impedance corresponding to each set of the current signals and the voltage signals may have different weights. In an embodiment, values of the weight of the transient impedance corresponding to each set of the current signals and the voltage signals may depend on the efficiency / performance of the transient impedance. In an embodiment, the transient impedance having the range of 52 to 48 may have a higher weight value than the transient impedance having a value greater than 52 to 48 and smaller than 52 to 48 . Further, the transient impedance having a value that suddenly spikes in a predetermined time may have a lesser weight value, without departing from the scope of the present disclosure. This configuration helps in monitoring and removing redundant values during the filtering of the transient impedance . Further, the filtering unit 210, after the filtering, generates the plurality of output values corresponding to the filtered set of the current signals and the voltage signals.

[0075] Further, the transformation unit 212 may be adapted to receive the plurality of output values corresponding to the filtered set of the current signals and the voltage signals from the filtering unit 210 to decrease the dimension of the plurality of output values and supply to the detection unit 214. This configuration ensures efficient transferring of the plurality of output values to the detection unit 214.

[0076] At block 406, the method includes processing, by the detection unit 214, the plurality of output values to identify the pattern using the supervised machine-learning model. The detection unit 214 is adapted to process the plurality of output values to identify the pattern using the supervised machine learning model.

[0077] According to the present disclosure, the training of the supervised machine learning model may be performed prior to the implementation of the method 400 An exemplary method of training the supervised machine learning model may be explained with reference to blocks 502 to 508 provided in Figure 5.

[0078] At block 502, the method includes receiving, by the training unit 216, the plurality of set of input signals corresponding to the different types of power sources 106, wherein a set of input signals corresponds to a type of power source. The training unit 216 may be adapted to receive the plurality of set of input signals corresponding to the different types of power sources 106. The set of input signals corresponds to the type of power source.

[0079] At block 504, the method includes filtering, by the training unit 216, each set of the input signals to generate a set of filtered values. The training unit 216 may be adapted to filter each set of the input signals, to generate the set of filtered values. For instance, each set of the input signals may be cleaned and filtered to remove redundant values. Further, each set of the input signals is standardized to generate the set of filtered values. Further, a feature selection or an extraction technique may be implemented on each set of the filtered values to reduce dimensionality of each set of the filtered values or transform the set of filtered values into a desired format.

[0080] Further, several different hyper-parameters of the supervised machine learning model need to be set before the training. The different hyper-parameters include a regularization parameter (C), which controls a trade-off between maximizing a margin and minimizing classification errors, and the kernel parameters (if applicable). Hyper-parameter tuning involves selecting the optimal values to achieve the best model performance.

[0081] At block 506, the method includes training, by the training unit 216, the supervised machine learning model to deduce a pattern from the set of filtered values. The training unit 216 may be adapted to train the supervised machine learning model to deduce the pattern from the set of filtered values.

[0082] In an embodiment, the supervised machine learning model may be trained prior to deployment of the supervised machine learning model to identify the pattern, where, the plurality of input signals corresponding to the different types of power sources 106 may be stored in a data repository of the supervised machine learning model, without departing from the scope of the present disclosure. In an embodiment, the electric load 108, for example, the cellular tower, receiving the power from the different power sources remains same, without departing from the scope of the present disclosure. In an embodiment, the supervised machine learning model may be trained for a predetermined time in a range of 25 sec. to 30 sec., without departing from the scope of the present disclosure.

[0083] Further, in another embodiment, the supervised machine learning model may be trained locally, where the supervised machine learning model may be trained in a real-time, after deployment of the supervised machine learning model in the detection unit 214 without departing from the scope of the present disclosure. In another embodiment, the electric load 108 receiving the power from the different power sources may differ, for example, from the cellular tower to a computer, without departing from the scope of the present disclosure. Further, the supervised machine learning model may be adapted to retrain in case of a change of the electric load 108, without departing from the scope of the present disclosure. For instance, the set of filtered values may be split into a training set values and a test set values. The training set values may be adapted to train the supervised machine learning model and the test set values may be adapted to test the supervised machine learning model. Further, the set of filtered values may be scaled to a standard range. Further, the supervised machine learning model may be trained by the training unit 216 by finding an optimal hyperplane that separates each of the set of filtered values, that is, the training set values, belonging to the different types of power sources 106 (as shown in Figure 6).

[0084] The supervised machine learning model may include different kinds of kernel functions to transform each set of the filtered values into a higher-dimensional feature space, which may be easier to find a separating hyperplane. Common kernel functions include linear, polynomial, Gaussian (RBF), and sigmoid. The hyperplane may be selected to maximize the margin, i.e., the distance between the hyperplane and the nearest data points of each class. The training process involves solving a constrained optimization problem using techniques such as quadratic programming. Further, each of the training set values may be assigned a target variable or a label that may represent the pattern to be deduced. Thus, the training unit 216 may be adapted to train the supervised machine learning model to deduce the pattern from the set of filtered values.

[0085] At block 508, the method includes associating, by the training unit 216, the deduced pattern with the type of power source. The training unit 216 may be adapted to associate the deduced pattern with the type of power source.

[0086] Further, when the supervised machine learning model may be trained, the supervised machine learning model may be evaluated using the test set values to determine efficiency of the supervised machine learning model. Further, the supervised machine learning model, after the evaluation, may be configured to be used in the PSIS 100.

[0087] Thus, once the pattern may be identified using the supervised machine learning model, the model can be deployed to perform the steps at block 408 in Figure 4.

[0088] At the block 408, the method includes comparing, by the detection unit 214, the identified pattern with a set of predetermined patterns corresponding to the different types of power sources 106 to identify the type of power source. The detection unit 214 may be adapted to compare the identified pattern with the set of previously learned patterns by the supervised machine learning model corresponding to the different types of power sources 106 to identify the type of power source.

[0089] According to the present disclosure, the detection unit 214 may be adapted to identify the type of power source locally, where the supervised machine learning model may be trained in the real-time, after deployment of the supervised machine learning model in the detection unit 214 and the PSIS 100 may be the standalone system, without departing from the scope of the present disclosure. In another embodiment, the detection unit 214 may be adapted to identify the type of the power source remotely, where the supervised machine learning model may be trained prior to deployment of the supervised machine learning model in the detection unit 214, and also, the training unit 216 and the detection unit 214 may be the cloud-server unit, without departing from the scope of the present disclosure.

[0090] In one example, each of the plurality of outputs of the fdtering unit 210 may be passed to the detection unit 214. Real-time processing of each of the plurality of outputs involves the transformation unit 212 having a linear discriminant analysis (LDA). The LDA may be adapted to transform the plurality of outputs into a lower-dimensional plurality of outputs. Further, the lower dimensional plurality of outputs may be provided to the detection unit 214 for further processing through the supervised machine learning model. In an embodiment, the supervised machine learning model may be a filtered vector machine algorithm, without departing from the scope of the present disclosure.

[0091] Thus, the detection unit 214 may be adapted to accurately compare the identified pattern with the set of a previously learned pattern by the supervised machine learning -model corresponding to the different types of power sources 106 to identify the type of power source.

[0092] Further, the plurality of output values and the filtered set of the current signals and the voltage signals processed by the trained supervised machine learning model by implementing the method shown in Figure 4 may also be subsequently used to train the same supervised machine learning model by implementing the method shown in Figure 5 to further enhance the accuracy of the supervised machine learning model. This type of training enables the real time training after the deployment of the supervised machine learning model. Further, the convergence analysis of the method as disclosed above followed by the supervised machine -learning module may be segregated into proving three consecutive convergences, for the filtering unit 210, the transformation unit 212, and detection unit 214 having the supervised machine learning model. The filtering unit 210 may be a bounded-input-bounded-output stable. Therefore, since the energy of the plurality of input signals may be bounded, the plurality of outputs generated by the filtering unit 210 converges with uniqueness . Further, the convergence of transforming unit 212 having the LDA depends upon the structure of an optimization function. The convergence analysis ensures optimal working of the PSIS 100.

[0093] As the optimization function may be smooth, bounded, and convex, thus the convergence of the transforming unit 212 having the LDA may be established. Similarly, based on the optimal hyperplane, the supervised machine learning model may be converging to a unique optimum. Therefore, the supervised machine learning model provided in the detection unit 214 may be convergent.

[0094] Further, a laboratory test set-up of the PSIS 100 with respect to the different types of power sources 106, is executed in a laboratory space. The result of the laboratory test set-up may be provided in the form of the graphical representations in conjunction with Figures 7A to 7C as explained in subsequent paragraphs.

[0095] Figure 7A illustrates a graph showing a precision versus a training length of the PSIS 100, according to an embodiment of the present disclosure. The graph provides the precision of the supervised machine learning module of the detection unit 214 of the PSIS 100. The precision obtained by the supervised machine learning model increases with an increase in the training length provided by the training unit 216. However, the precision starts decreasing once a maximum value of the precision may be reached. Further, after reaching the maximum value, the precision either saturates or decreases based on the sampling frequency. This decrease in precision may be a consequence of over-training the supervised machine learning module at the higher training lengths. Further, the optimal training length may be 500 samples at a sampling rate of 700 sps, resulting in a precision of > 0.99 precision.

[0096] Figure 7B illustrates a graph showing an accuracy versus a bandwidth of a kernel function of the PSIS 100, according to an embodiment of the present disclosure. The graph provided indicates that a classification accuracy increases with an increase in a kernel bandwidth of a radial basis function for all training lengths. As shown in the graph, at the optimal training length of 500 samples, the kernel bandwidth of y ~ 1.5 yields 100% accuracy. A minor decrease in classification accuracy may be observed around this kernel bandwidth. Therefore, the graph Indicates that the PSIS 100 performs efficiently at the kernel bandwidth of y ~ 1.5, resulting in the classification accuracy of « 99.2%. During the simulation, the supervised machine learning model of the detection unit 214 exhibits optimal performance at a lag length of 2 samples with filter coefficients specified by [0.7, 0.3],

[0097] Figure 8A illustrates a graph showing an accuracy versus a training length at a predetermined sampling frequency of the PSIS 100, according to an embodiment of the present disclosure. The graph shows that the PSIS 100 as disclosed in the present disclosure achieves nearly 100% accuracy in 500 training samples, which corresponds to approximately 0.5 seconds of recorded data at a sampling rate of 1000 sps, due to optimal data filtering. In contrast, the accuracy of all the other existing systems of a state of the art increases with the increase in the training length. Therefore, the PSIS 100 achieves optimal performance in a lesser fraction of time, resulting in a 76.47% gain in the accuracy of classification. As a result, the points of switchovers may be identified with a high degree of accuracy.

[0098] Figure 8B illustrates a graph showing a sensitivity versus a sampling frequency at the training length of the predetermined samples of the PSIS 100, according to an embodiment of the present disclosure. The graph shows that the PSIS 100 works with close to unit sensitivity, indicating that the PSIS 100 may be able to repeatedly classify the points in the correct classes with high accuracy. Furthermore, the sensitivity increases with the sampling frequency, resulting from the training of the supervised machine learning model done on multiple data points. In contrast, all the other existing systems of state-of-the-art operate below 0.6 sensitivity due to a flattened DC signature at the output end of the DC power management module. However, the present disclosure includes the filtering unit 210 to appropriately identify the relevant features, leading to an improved performance in the source classification. Thus, on comparative analysis, the PSIS 100 achieves 66.67% gain in sensitivity over the state-of-the- art.

[0099] Figure 8C illustrates a graph showing an accuracy versus a measurement noise at the training length of the predetermined sample, and the kernel bandwidth of the PSIS 100, according to an embodiment of the present disclosure. The graph shows that a performance of the existing system of state of the art, for example, a decision tree and a SVM classifiers degrade significantly as the noise in the measurement data increases, which is mainly due to the absence of any finite impulse response filter. However, the PSIS 100 employs the moving average filtering unit 210, ensuring robustness to the increase in the noise and operate with accuracy of approximately to 100%. Therefore, the PSIS 100 provides an accurate and robust means of classifying the input sources of power, e., the type of power source from the different types of power sources 106, based on a load-end DC signature.

[0100] Further, a field test-set up of the PSIS 100 with respect to the different type of power sources 106, is executed in a field space. For the field test-set up, the different types of power sources 106 incorporated may be a power grid, a battery, and a diesel generator. Further, the result of the field test-set up may be provided in the form of the graphical representations as explained in subsequent paragraphs in conjunction with Figures 9A and 9B.

[0101] Figure 9A illustrates a graph showing a type of power source from different types of power sources 106 and a switchover identification in voltage measurements of the PSIS 100, according to an embodiment of the present disclosure. Figure 9B illustrates a graph showing a type of power source from different types of power sources 106 and a switchover identification in current measurements of the PSIS 100, according to an embodiment of the present disclosure. The graphs show the load-end DC current and the voltage dataset generated by recording the voltage and current characteristic of each source roughly for 5 minutes each and finally with multiple switchovers generated between the power grid (G), battery (B), and diesel generator (D) power sources. Two sets of individual source signatures may be noted, resulting in 10 minutes of data per source. Thus, a total of 30 minutes of data samples may be recorded including all three sources. To record all possible switchovers, the data may be recorded in the pattern G-D-B-D-G-B-G with an arbitrarily random duration, between 1-2 minutes, spent per source. Therefore, an approximate duration of 14-15 minutes may be considered to record source switchovers. In total, an aggregate data length of 45 minutes may be collected for one site. The PSIM technology was verified in a similar manner for two other sites as well, whose loading characteristics are defined in Table 1 :

[0102] Table 1

[0103] Figure 10A illustrates a graph showing an accuracy versus a training length of the PSIS 100 in the laboratory set-up and the field set-up, according to an embodiment of the present disclosure. The graph shows that the PSIS 100 performs with more than 99% accuracy in both the laboratory set-up and the field test set-up. Further, in the field set-up, the PSIS 100 requires an even lesser number of samples for training than under the laboratory set-up, to achieve more than 99% accuracy. This is a result of a controlled laboratory environment, which allowed a generation of a more accurate DC voltage profile. From a comparative study, in the field setup, the PSIS 100 achieves more than 99 % accuracy with 85.71% reduced training samples than required in the laboratory set-up, which corresponds to a data record time of 1 ms at a sampling frequency of 1000 sps.

[0104] Figure 10B illustrates a graph showing an accuracy versus a sampling frequency of the PSIS 100 in the laboratory set-up and the field test setup, according to an embodiment of the present disclosure. The graph shows in the field set-up, that the PSIS 100 achieves approximately 100% performance at a sampling frequency greater than or equal to 550. Therefore, from this comparative study of the PSIS 100 in the laboratory set-up and the field set-up, conclusion may be that the PSIS 100 achieves desired performance with almost 100% classification accuracy in actual field deployment conditions. Furthermore, this classification is achieved with a training time of approximately 1 ms. Figure IOC illustrates a graph identification of the type of source from the different types of sources 106 using the PSIS 100 in the laboratory set-up and the field test setup, according to an embodiment of the present disclosure. The graph depicts the source identification demonstrated by the PSIS 100 on the collective field dataset that contained multiple switchovers among the three sources. For instance, grid (G) is mapped with 1, battery (B) is mapped to 2, and the diesel generator source (D) is mapped to 3. The identification logic is applied as, if the classification accuracy may be greater than 99%, a 100% source identification accuracy is achieved. The main idea behind this logic is that 1% misclassification implies that 0.01 fraction of the samples collected in a second is misclassified. This translates to 10 misclassified samples, or a misclassification duration of 10 ms, which is extremely small to experience the source switchover. Therefore, a 99% correct classification may be used to identify a source correctly using the 990 correctly identified samples. Based on this understanding, the PSIS 100 provides a 100% accurate identification of the source labels. Therefore, the PSIS 100 may be able to provide 100% correct source identification as well as detection of source switchovers. This is a key requirement for successful field deployment of the PSIS 100.

[0105] As would be gathered, the present disclosure ensures a configuration of the PSIS 100 to identify the type of power source from the different types of power sources 106 with the assistance of the supervised machine learning model. The PSIS 100 may be disposed at the end of the load bus 102, that is, a DC load bus. The identification of the type of source power ensures that the PSIS 100 may be adapted to detect the type of power source which may be generating the harmonics. This configuration maintains the life and working of the electric load 108. Further, if the type of the power source may get damaged, the PSIS 100 may be adapted to detect the switchover from the type of power source to another type of power source and thus maintains a continuous supply of the power to the electric load 108 through the DC power management. Further, the configuration ensures detection of switchover to the different types of power sources 106 based on the identified harmonics pattern by a power source in the electric load 108 through the load bus 102. This configuration eliminates impact of the harmonics in the electric load 108 through the load bus 102, by ensuring that necessary actions are implemented to improve harmonic profile of a given input power source. The present configuration also restricts accessing a power management system, i.e., provides leverage to the electric load 108, where the electric load 108 may identify activity of any input power source, i.e., instants of switchover, source label, and all other details that can be generated as a byproduct, without having to tap into a system which could be outside jurisdiction of the electric load 108. The configuration as disclosed ensures efficient deduction of a DC load pattern.

[0106] The PSIS 100 works with more than approximately 76% gain in the accuracy and more than approximately 66% gain in the sensitivity over the state-of-the-art and is highly robust to measurement noise. Further, with optimal settings of the filter coefficients and the sampling frequency, the PSIS 100 attains more than approximately 99% classification accuracy with around 100% source identification performance. The PSIS 100 may be additionally used to identify load switchover in AC and DC grids, fault and islanding detection, harmonic identification in pulsed load systems, and classification of disturbance sources in smart grid networks. Further, the transmission of the identified pattern in the CDMS 110 ensures providing information, for example, the harmonic source identification, the load distribution of each of the different types of power sources 106, etc. This provides a flexibility to select the optimum type of power source from the different types of power sources 106, thus resulting in cost-effective power supply to the electric load, etc.

[0107] While specific language has been used to describe the present subject matter, any limitations arising on account thereto, are not intended. As would be apparent to a person in the art, various working modifications may be made to the method in order to implement the inventive concept as taught herein. The drawings and the foregoing description give examples of embodiments. Those skilled in the art will appreciate that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be split into multiple functional elements. Elements from one embodiment may be added to another embodiment.

Claims

We Claim:

1. A power source identification system (PSIS) (100) for detecting a type of power source connected to an electric load (108) via a load bus (102), comprising: a sensor unit (208) connected to the load bus (102) and adapted to measure amperage and voltage of electric current flowing through the load bus (102) to generate a set of current signals and voltage signals within a predetermined time interval, respectively; a filtering unit (210) adapted to process each set of the current signals and the voltage signals to filter noise therein and generate a plurality of output values corresponding to a filtered set of the current signals and the voltage signals; and a detection unit (214) adapted to: process the plurality of output values to identify a pattern using a supervised machine learning model; and compare the identified pattern with a set of previously learned patterns by the supervised machine learning model corresponding to different types of power sources (106) to identify the type of power source.

2. The PSIS (100) as claimed in claim 1, wherein the sensor unit (208) comprises: a voltage sensor connected across an end of the load bus (102) and adapted to measure voltage of the electric current; and a current sensing unit adapted to connect to the load bus (102) in a non-invasive manner to measure the electric current.

3. The PSIS (100) as claimed in claim 1, comprising: a training unit (216) adapted to: receive a plurality of set of input signals corresponding to the different type of power sources, wherein a set of input signals correspond to a type of power source; filter, each set of input signals, to generate, a set of filtered values; train the supervised machine learning model to deduce a pattern from the set of filtered values; and associate the deduced pattern with the type of power source.

4. The PSIS (100) as claimed in claim 1, wherein the sensor unit (208) samples the amperage and the voltage of the electric current, flowing through the load bus (102), at a predetermined sampling rate of up to 1000 samples per second to generate the set of current signals and the voltage signals within the predetermined time interval, and the filtering unit (210) is a moving average filtering unit adapted to process each set of the current signals and the voltage signals to filter the noise therein and generate the plurality of output values corresponding to the filtered set of the current signals and the voltage signals.

5. The PSIS (100) as claimed in claim 1, comprising a transformation unit (212) adapted to receive the plurality of output values, corresponding to the filtered set of the current signals and the voltage signals from the filtering unit (210), to decrease dimension of the plurality of output values and supply to the detection unit.

6. The PSIS (100) as claimed in claim 1, wherein the detection unit (214) is adapted to communicate the identified patterns to a central data management system (110) through one of a wired or a wireless network.

7. The PSIS (100) as claimed in claim 1, wherein the sensor unit (208), filtering unit (210), and the detection unit (214) are disposed in proximity to the load bus (102).

8. The PSIS (100) as claimed in claim 1, wherein the sensor unit (208), the filtering unit (210) are adapted to be disposed in proximity to the load bus (102) and the detection unit (214) is a cloud server unit wherein the detection unit (214) is adapted to be communicatively coupled with the sensor unit and the filtering unit.

9. A method (400) for detecting a type of power source connected to an electric load (108) via a load bus (102) through a power source identification system (PSIS) (100), comprising: sensing (402), by a sensor unit (208), an electric current and measuring an amperage and voltage of the electric current flowing through a load bus (102) to generatea set of current signals and voltage signals within a predetermined time interval, respectively; filtering (404), by a filtering unit (210), noise from each set of the current signals and the voltage signals therein and generating, by the fdtering unit (210), a plurality of output values corresponding to a fdtered set of current signals and voltage signals; processing (406), by a detection unit (214), the plurality of output values to identify a pattern using a supervised machine learning model; and comparing (408), by the detection unit (214), the identified pattern with a set of predetermined patterns corresponding to different types of power sources (106) to identify the type of power source.

10. The method (400) as claimed in claim 9, wherein the method (400) further comprises: receiving (502), by a training unit (216), a plurality of set of input signals corresponding to the different type of power sources, wherein a set of input signal corresponds to a type of power source; filtering (504), by the training unit (216), each set of input signals, to generate, a set of filtered values; training (506), by the training unit (216), the supervised machine learning model to deduce a pattern from the set of filtered values; and associating (508), by the training unit (216), the deduced pattern with the type of power source.

Images