Blind source separation transfer learning monitoring method, power grid system, device and storage medium

Abstract

The invention discloses a blind source separation transfer learning monitoring method, a power grid system, a blind source separation transfer learning monitoring device and a storage medium, and the method comprises the steps: obtaining base current data to construct a training data set, firstly controlling each circuit breaker to be conducted for steady-state operation, sequentially controlling the on-off action of each circuit breaker to obtain current waveform base signals collected by a current transformer in each on-off process to construct a training data set; obtaining an initial separation matrix by using the training data set and the branch data set, and establishing a blind source separation model; iteratively executing a transfer learning step for multiple times: acquiring a real-time main circuit current; obtaining an estimated current waveform signal of each execution branch based on a blind source separation model according to the real-time main circuit current; obtaining the on-off action of each execution branch, and analyzing the variation of the separation matrix; iteratively calculating a target separation matrix, and updating a blind source separation model at the next moment by using the target separation matrix; according to the design, the monitoring structure is simplified, the operation cost is reduced, and the diagnosis accuracy is optimized.

Description

Technical Field

[0001] This invention relates to the technical field of power system monitoring and diagnostic methods, and in particular to a blind source separation transfer learning monitoring method, as well as a power grid system, device, and storage medium. Background Technology

[0002] High-voltage circuit breakers are widely used in power distribution systems. As a core protection device, the operating status of the circuit breaker directly determines the safety and reliability of the system.

[0003] The existing methods for monitoring the execution branch where high-voltage circuit breakers are located mostly involve installing current or voltage sensors inside the circuit breaker. Installation and maintenance require shutdown, which greatly increases operating costs and affects the continuous operation of the power system. Summary of the Invention

[0004] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a blind source separation transfer learning monitoring method, as well as a power grid system, device, and storage medium, simplifying the monitoring structure, reducing operating costs, and optimizing diagnostic accuracy.

[0005] According to a first aspect of the present invention, a blind source separation transfer learning monitoring method is applied to a power grid system. The power grid system includes multiple execution branches, a main busbar, current transformers, and a control module. Each execution branch is equipped with a circuit breaker to control the on / off state of the execution branch. Each execution branch is connected to the main busbar. The current transformers detect the main current of the main busbar. The control module is connected to the current transformers. The method is characterized in that the control module executes a blind source separation transfer learning monitoring method, which includes: To acquire base current data to construct a training dataset, firstly, control each circuit breaker to operate in a steady state while conducting, and then sequentially control the switching actions of each circuit breaker to obtain the base current waveform signals collected by the current transformer during each switching process. The training dataset is constructed from the base current waveform signals. An initial separation matrix is ​​obtained using the training dataset and the branch dataset, and a blind source separation model is established based on the initial separation matrix. The branch dataset includes the branch current waveform signal when the circuit breaker in each branch is switched on or off. The power grid system is put into operation and multiple transfer learning steps are executed iteratively. These transfer learning steps include: Obtain the real-time main circuit current; The estimated current waveform signals of each execution branch are obtained based on the real-time main current and the blind source separation model. The switching actions of each execution branch at that moment are obtained, and the change in the separation matrix is ​​analyzed based on the estimated current waveform signal of each execution branch, the training dataset, and the deviation value of the branch dataset. The target separation matrix is ​​calculated iteratively using the change in the separation matrix, and the blind source separation model is updated using the target separation matrix to obtain the estimated current waveform signal of each execution branch at the next moment. The blind source separation model after the transfer learning step is used to process the real-time main current to obtain the estimated current waveform signal of each execution branch.

[0006] The blind source separation transfer learning monitoring method according to embodiments of the present invention has at least the following beneficial effects: This invention relates to a blind source separation transfer learning monitoring method. Before each circuit breaker is put into use in the power grid system, the branch current waveform signal generated by the operation of the branch to which the circuit breaker is located can be detected individually, thereby forming a branch dataset. After being put into use in the power grid system, current transformers collect the main current of the bus main circuit into which each branch is connected. In acquiring the base current data, each circuit breaker is first controlled to operate in a steady state, and then the on and off actions of each circuit breaker are controlled sequentially to obtain the base current waveform signal collected by the current transformer during each on and off process. Thus, each base current waveform signal is associated with the corresponding branch current waveform signal. By constructing the training dataset and the branch dataset using each base current waveform signal, the initial separation matrix can be obtained, and the root... A blind source separation model is established based on the initial separation matrix. Then, the blind source separation model is applied to the actual operation of the power grid system to perform a transfer learning step to obtain the real-time main current. At the moment of obtaining the real-time main current, the switching actions of each execution branch are accompanied by the switching actions of each execution branch. Based on the estimated current waveform signal of each execution branch, the training dataset, and the branch dataset, the deviation value of the waveform signal can be obtained, thereby analyzing the change of the separation matrix. In multiple transfer learning steps, the target separation matrix is ​​continuously iterated, and finally a better blind source separation model is formed. The estimated current waveform signal of each execution branch can be obtained by processing the real-time main current using the blind source separation model. This design simplifies the monitoring structure, reduces operating costs, and optimizes the accuracy of diagnosis.

[0007] According to some embodiments of the present invention, the blind source separation model is as follows: ; in, Let be the estimated current waveform signal of each execution branch at time t. Let be the target separation matrix at time t. Let t be the main current obtained at time t.

[0008] According to some embodiments of the present invention, the process of deriving the initial separation matrix using the training dataset and the branch dataset includes: In the training dataset and the branch dataset, the base signal of the current waveform when each circuit breaker is switched on and off and the corresponding branch current waveform signal are calculated to obtain each initial separation element. The initial separation matrix is ​​obtained by transposing each of the separation elements into vectors.

[0009] According to some embodiments of the present invention, the step of analyzing the change in the separation matrix based on the estimated current waveform signal of each execution branch, the training dataset, and the deviation value of the branch dataset includes: Each learning fusion separation element is calculated based on the estimated current waveform signal of each execution branch and the corresponding branch current waveform signal. The change in the separation matrix is ​​calculated by comparing the deviation values ​​of each learned fusion separation element with the corresponding initial separation element.

[0010] According to some embodiments of the present invention, the blind source separation model that iteratively calculates the target separation matrix using the change in the separation matrix and updates the estimated current waveform signal of each execution branch at the next moment using the target separation matrix includes: ; in, Let be the target separation matrix at time t. Initial separation moment Formation, This represents the matrix change during the k-th iteration.

[0011] According to some embodiments of the present invention, the blind source separation transfer learning monitoring method includes: setting an iteration threshold, and when the number of iterations reaches the iteration threshold, the training of the blind source separation model is completed.

[0012] According to some embodiments of the present invention, when the change in the separation matrix is ​​less than the threshold value in the last transfer learning step, the transfer learning step is terminated.

[0013] According to a second aspect of the present invention, a power grid system includes multiple execution branches, a main bus, current transformers, and a control module. Each execution branch is equipped with a circuit breaker to control the opening and closing of the execution branch. Each execution branch is connected to the main bus. The current transformer detects the main current of the main bus. The control module is connected to the current transformer. The control module executes the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0014] The power grid system according to embodiments of the present invention has at least the following beneficial effects: In the power grid system of this invention, the control module executes the blind source separation transfer learning monitoring method disclosed in any of the above embodiments to obtain a better blind source separation model, and then uses the blind source separation model to process the real-time main current to obtain the estimated current waveform signal of each execution branch. This design simplifies the monitoring structure, reduces operating costs, and optimizes the accuracy of diagnosis.

[0015] According to a third aspect of the present invention, the control device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0016] According to a fourth aspect of the present invention, a computer-readable storage medium stores a computer program, characterized in that, when the computer program is executed by a processor, it implements the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a schematic diagram of the principle structure of one embodiment of the power grid system of the present invention; Figure 2 This is a main flowchart of one embodiment of the blind source separation transfer learning monitoring method of the present invention; Figure 3 This is a flowchart of step S530 of one embodiment of the blind source separation transfer learning monitoring method of the present invention; Figure 4 This is a schematic diagram of the control device of the present invention in one embodiment.

[0019] Figure label: Execution branch 100; circuit breaker 110; main bus 200; current transformer 300; control module 400; processor 610; memory 620; input / output interface 630; communication interface 640; bus 650. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0021] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0022] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than", "less than", "exceeding" are understood to exclude the number itself, and "above", "below", "within" are understood to include the number itself.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0024] like Figures 1 to 3 As shown, the blind source separation transfer learning monitoring method according to the first aspect of the present invention is applied to a power grid system. The power grid system includes multiple execution branches 100, a main bus 200, a current transformer 300, and a control module 400. Each execution branch 100 is equipped with a circuit breaker 110 to control the on / off state of the execution branch 100. Each execution branch 100 is connected to the main bus 200. The current transformer 300 detects the main current of the main bus 200. The control module 400 is connected to the current transformer 300.

[0025] like Figure 1 As shown, the current transformer 300 is installed in the main bus 200, and the current of each execution branch 100 is fed into the main bus 200. The circuit breaker 110 is connected in series in the corresponding execution branch 100. The control module 400 includes a processor and its auxiliary circuits. The control module 400 obtains the main current through the current transformer 300.

[0026] The control module executes the blind source separation transfer learning monitoring method, which includes: S510. Obtain base current data to construct a training dataset. First, control each circuit breaker to operate in a steady state. Then, control the switching actions of each circuit breaker in sequence to obtain the base current waveform signal collected by the current transformer during each switching process. Construct a training dataset from each base current waveform signal. S520. The initial separation matrix is ​​obtained using the training dataset and the branch dataset, and a blind source separation model is established based on the initial separation matrix. The branch dataset includes the branch current waveform signal when the circuit breaker in each execution branch is turned on or off. S530, Put the power grid system into operation and iteratively execute multiple transfer learning steps, such as Figure 3 As shown, the transfer learning steps include: S531, Obtain the real-time main circuit current; S532. Based on the real-time main current and the blind source separation model, the estimated current waveform signal of each execution branch is obtained. S533. Obtain the on / off actions of each execution branch at this moment, and analyze the change in the separation matrix based on the estimated current waveform signal of each execution branch, the training dataset, and the deviation value of the branch dataset. S534. The target separation matrix is ​​calculated iteratively using the change of the separation matrix, and the blind source separation model for obtaining the estimated current waveform signal of each execution branch at the next moment is updated using the target separation matrix. S540. The blind source separation model after the transfer learning step is completed is used to process the real-time main current to obtain the estimated current waveform signal of each execution branch.

[0027] Understandably, before each circuit breaker is put into use in the power grid system, the branch current waveform signal generated by the operation of the branch to which the circuit breaker is located can be detected individually, thus forming a branch dataset. After being put into use in the power grid system, the current transformer collects the main current of the bus main circuit into which each branch is connected. In acquiring the base current data, each circuit breaker is first controlled to operate in a steady state, and then the on and off operations of each circuit breaker are controlled sequentially to obtain the base current waveform signal collected by the current transformer during each on and off process. Thus, each base current waveform signal is associated with the corresponding branch current waveform signal. By constructing the training dataset and the branch dataset using each base current waveform signal, the initial separation matrix can be obtained, and based on the initial separation... A blind source separation model is established using a matrix, and then applied to the actual operation of the power grid system to perform a transfer learning step. This process acquires the real-time main current. During the acquisition of the real-time main current, the switching actions of each execution branch are considered. Based on the estimated current waveform signals of each execution branch, the training dataset, and the branch dataset, the deviation of the waveform signals can be determined, thereby analyzing the change in the separation matrix. Through multiple transfer learning steps, the target separation matrix is ​​continuously iterated, ultimately forming a superior blind source separation model. This model can then be used to process the real-time main current to obtain the estimated current waveform signals of each execution branch. This design simplifies the monitoring structure, reduces operating costs, and optimizes diagnostic accuracy.

[0028] In some embodiments of the present invention, the blind source separation model is as follows: ; in, Let be the estimated current waveform signal of each execution branch at time t. Let be the target separation matrix at time t. Let t be the main current obtained at time t.

[0029] At any given time, the acquired main current is input into the blind source separation model. By calculating the separation matrix with the target, the estimated current waveform signal of each execution branch can be separated, which can be used as a reference for the system or staff to diagnose the stable and safe operation of the execution branch.

[0030] In some embodiments of the present invention, the process of deriving the initial separation matrix using the training dataset and the branch dataset includes: In the training dataset and the branch dataset, the base signal of the current waveform when each circuit breaker is switched on and off and the corresponding branch current waveform signal are calculated to obtain each initial separation element. The initial separation matrix is ​​obtained by transposing each of the separation elements into vectors.

[0031] It is understandable that the base signal of the current waveform during the switching action of each circuit breaker is fused with the waveforms of other circuit breakers during steady-state operation. The branch current waveform signal during the operation of this circuit breaker and the branch current waveform signals during the steady-state operation of other circuit breakers can form a separate branch data matrix. Substituting the base signal of the current waveform and the separate branch data matrix into the blind source separation model yields the initial separation elements, which are... , where m is the m-th circuit breaker.

[0032] Then, the vector transpose of each separating element is used to obtain the initial separating matrix, i.e. ; Where T is the vector transpose function. This is the initial separation matrix.

[0033] In some embodiments of the present invention, when obtaining each initial separation element, the initial separation element can also be optimized to maximize non-Gaussianity.

[0034] The following model is used for optimization: ; in, Find the argument of a complex number. This is the base signal of the current waveform when the m-th circuit breaker operates.

[0035] G() is a non-Gaussianity metric function, i.e.: ; in, Negentropy is the negative entropy function, used to reflect the degree of non-Gaussianity. For mathematical expectation, Let be a Gaussian random variable used for reference.

[0036] In some embodiments of the present invention, the step of analyzing the change in the separation matrix based on the estimated current waveform signal of each execution branch, the training dataset, and the deviation value of the branch dataset includes: Each learning fusion separation element is calculated based on the estimated current waveform signal of each execution branch and the corresponding branch current waveform signal. The change in the separation matrix is ​​calculated by comparing the deviation values ​​of each learned fusion separation element with the corresponding initial separation element.

[0037] Understandably, the switching actions of the circuit breakers in each execution branch are controlled by the control module. The system can know the switching actions of the circuit breakers. When any one or more circuit breakers switch on or off while other circuit breakers are operating in a steady state, the system can form an action branch data matrix based on the current waveform signals of the corresponding circuit breaker branches. By substituting the estimated current waveform signals of each execution branch and the action branch data matrix into the blind source separation model, the learning fusion separation elements can be obtained.

[0038] The target separation matrix is ​​iteratively updated based on the change in the separation matrix obtained from the deviation between each learned fusion separation element and the corresponding initial separation element.

[0039] In some embodiments of the present invention, the blind source separation model that iteratively calculates the target separation matrix using the change in the separation matrix and updates the estimated current waveform signal for each execution branch at the next moment using the target separation matrix includes: ; in, Let be the target separation matrix at time t. The initial separation matrix ， This represents the matrix change during the k-th iteration.

[0040] It is understandable that the target separation matrix from the previous time step is iteratively updated in each transfer learning step.

[0041] Specifically, during each iteration update, it is also possible to... Gradient fine-tuning is performed to prevent deviation from the original model, i.e.: ; in, Let be the target separation matrix at time t. The updated target separation matrix, The learning rate can be set by staff according to actual needs. For loss function, The gradient of the loss function with respect to the separation matrix. This is a regularization parameter that can be set by staff according to actual needs to prevent overfitting and weight drift.

[0042] Domain-specific adaptive regularization: ; in, It is the L2 norm. For historical templates / prior circuit breaker operating characteristics, The weights for the distribution consistency constraint can be set by the staff. As a measure of the difference in the distribution of the source signals, we have: ; in, These represent the number of samples in the source domain and the target domain, respectively. is the mapping function of the sample in the kernel space.

[0043] From the above, we can derive the target separation matrix function for fine-tuning as follows: .

[0044] In some embodiments of the present invention, the blind source separation transfer learning monitoring method includes: setting an iteration threshold, and when the number of iterations reaches the iteration threshold, the training of the blind source separation model is completed.

[0045] The iteration threshold can be set as a specific number of iterations or a time value. The control module acquires the main current at a fixed acquisition cycle. Setting the time value can constrain the number of iterations. The iteration threshold is used to limit the number of iterations. When the iteration threshold is reached, the training of the blind source separation model is completed, thereby preventing overfitting and ensuring that the blind source separation model can be put into use in the power grid in a timely manner.

[0046] In some embodiments of the present invention, the transfer learning step is terminated when the change in the separation matrix is ​​less than the threshold value in the last transfer learning step.

[0047] It is understandable that when the change in the separation matrix is ​​less than the threshold, it means that the blind source separation model is within a reasonable range, which can ensure reasonable analytical accuracy and can be used in the power grid system. Specifically, the threshold can be set to 10% or 5%, etc., which is selected by the staff. If the change in the separation matrix is ​​still greater than the threshold in the last transfer learning step, then steps S510 to S530 are repeated.

[0048] According to a second aspect of the present invention, a power grid system, such as Figure 1As shown, the system includes multiple execution branches 100, a main bus 200, a current transformer 300, and a control module 400. Each execution branch 100 is equipped with a circuit breaker 110 to control the on / off state of the execution branch 100. Each execution branch 100 is connected to the main bus 200. The current transformer 300 detects the main current of the main bus 200. The control module 400 is connected to the current transformer 300. The control module 400 executes the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0049] Specifically, the power grid system may employ devices that implement the blind source separation transfer learning monitoring method disclosed in any of the above embodiments, which will not be elaborated here.

[0050] In the power grid system of the present invention, the control module 400 executes the blind source separation transfer learning monitoring method disclosed in any of the above embodiments to obtain a better blind source separation model, and then uses the blind source separation model to process the real-time main current to obtain the estimated current waveform signal of each execution branch 100. This design simplifies the monitoring structure, reduces operating costs, and optimizes the accuracy of diagnosis.

[0051] According to a third aspect of the present invention, the control device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0052] The control device can be any intelligent terminal, including a central computer, a remote equipment terminal computer, or any other intelligent terminal.

[0053] like Figure 4 As shown, Figure 4 The hardware structure of a control device according to another embodiment is also illustrated. The control device includes: The processor 610 can be implemented using a general-purpose central processing unit (CPU), a microprocessor 610, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 620 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 620 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 620 and is called and executed by the processor 610 using the blind source separation transfer learning monitoring method of the embodiments of this application. The input / output interface 630 is used to realize information input and output; The communication interface 640 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). The bus 650 transmits information between various components of the device (such as processor 610, memory 620, input / output interface 630 and communication interface 640), and can also be connected to the smart Internet of Things. The processor 610, memory 620, input / output interface 630 and communication interface 640 are connected to each other within the device via bus 650.

[0054] According to a fourth aspect of the present invention, a computer-readable storage medium stores a computer program, characterized in that, when executed by a processor, the computer program implements the blind source separation transfer learning monitoring method disclosed in any of the above embodiments.

[0055] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0056] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0057] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0058] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0059] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0060] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0061] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

[0062] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A blind source separation transfer learning monitoring method, applied to a power grid system, wherein the power grid system includes multiple execution branches, a main bus, current transformers, and a control module. Each execution branch is equipped with a circuit breaker to control the on / off state of the execution branch. Each execution branch is connected to the main bus. The current transformers detect the main current of the main bus. The control module is connected to the current transformers. The method is characterized by... The control module executes the blind source separation transfer learning monitoring method, which includes: To acquire base current data to construct a training dataset, firstly, control each circuit breaker to operate in a steady state while conducting, and then sequentially control the switching actions of each circuit breaker to obtain the base current waveform signals collected by the current transformer during each switching process. The training dataset is constructed from the base current waveform signals. An initial separation matrix is ​​obtained using the training dataset and the branch dataset, and a blind source separation model is established based on the initial separation matrix. The branch dataset includes the branch current waveform signal when the circuit breaker in each branch is switched on or off. The power grid system is put into operation and multiple transfer learning steps are executed iteratively. These transfer learning steps include: Obtain the real-time main circuit current; The estimated current waveform signals of each execution branch are obtained based on the real-time main current and the blind source separation model. The switching actions of each execution branch at that moment are obtained, and the change in the separation matrix is ​​analyzed based on the estimated current waveform signal of each execution branch, the training dataset, and the deviation value of the branch dataset. The target separation matrix is ​​calculated iteratively using the change in the separation matrix, and the blind source separation model is updated using the target separation matrix to obtain the estimated current waveform signal of each execution branch at the next moment. The blind source separation model after the transfer learning step is used to process the real-time main current to obtain the estimated current waveform signal of each execution branch.

2. The blind source separation transfer learning monitoring method according to claim 1, characterized in that, The blind source separation model is as follows: ； in, Let be the estimated current waveform signal of each execution branch at time t. Let be the target separation matrix at time t. Let t be the main current obtained at time t.

3. The blind source separation transfer learning monitoring method according to claim 1, characterized in that, The initial separation matrix derived using the training dataset and the branch dataset includes: In the training dataset and the branch dataset, the base signal of the current waveform when each circuit breaker is switched on and off and the corresponding branch current waveform signal are calculated to obtain each initial separation element. The initial separation matrix is ​​obtained by transposing each of the separation elements into vectors.

4. The blind source separation transfer learning monitoring method according to claim 3, characterized in that, The analysis of the separation matrix change based on the estimated current waveform signals of each execution branch, the training dataset, and the deviation values ​​of the branch dataset includes: Each learning fusion separation element is calculated based on the estimated current waveform signal of each execution branch and the corresponding branch current waveform signal. The change in the separation matrix is ​​calculated by comparing the deviation values ​​of each learned fusion separation element with the corresponding initial separation element.

5. The blind source separation transfer learning monitoring method according to claim 4, characterized in that, The blind source separation model, which iteratively calculates the target separation matrix using the change in the separation matrix and updates the estimated current waveform signal for each execution branch at the next moment using the target separation matrix, includes: ； in, Let be the target separation matrix at time t. Initial separation moment Formation, This represents the matrix change during the k-th iteration.

6. The blind source separation transfer learning monitoring method according to claim 5, characterized in that, include: Set an iteration threshold; when the number of iterations reaches the threshold, the training of the blind source separation model is complete.

7. The blind source separation transfer learning monitoring method according to claim 6, characterized in that, The transfer learning process ends when the change in the separation matrix is ​​less than the threshold in the final transfer learning step.

8. A power grid system, characterized in that, It includes multiple execution branches, a main busbar, current transformers, and a control module. Each execution branch is equipped with a circuit breaker to control the on / off state of the execution branch. Each execution branch is connected to the main busbar. The current transformer detects the main current of the main busbar. The control module is connected to the current transformer. The control module executes the blind source separation transfer learning monitoring method as described in any one of claims 1 to 7.

9. A control device, characterized in that, The control device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the blind source separation transfer learning monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the blind source separation transfer learning monitoring method according to any one of claims 1 to 7.

Images