Transformer fault identification method and apparatus, and device and medium

Abstract

Provided are a transformer fault identification method and apparatus, and a device and a medium. The method comprises: during the process of optimizing a convolutional neural network model on the basis of an optimization algorithm, determining a weight decay term on the basis of a parameter value after last iteration; on the basis of the parameter value after last iteration and the weight decay term, determining a parameter value after the next iteration; in response to determining that the parameter value after the next iteration meets a preset condition, determining a corresponding target weight and target bias; on the basis of the target weight and the target bias, determining an optimized convolutional neural network model; and acquiring concentration data of dissolved gases in transformer oil, and processing the concentration data on the basis of the optimized convolutional neural network model to identify a transformer fault type.

Description

Transformer fault identification method, device, equipment and medium

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on July 31, 2023, with application number 202310958219.1, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the technical field of transformers, and for example, to a method, device, equipment, and medium for identifying transformer faults. Background Art

[0003] Accurate fault detection for oil-immersed transformers is a pressing technical challenge for those skilled in the art. Commonly used transformer fault diagnosis algorithms include support vector machines (SVMs), fuzzy logic (FL), and convolutional neural networks (CNNs). However, these machine learning models suffer from slow training convergence and low training accuracy.

[0004] Summary of the Invention

[0005] The present application provides a transformer fault identification method, apparatus, device and medium, which can improve the attenuation effect and the accuracy of model identification during the model optimization process.

[0006] According to one aspect of the present application, a transformer fault identification method is provided, the method comprising:

[0007] In the process of optimizing the convolutional neural network model based on the optimization algorithm, the weight decay term is determined according to the parameter value after the previous iteration;

[0008] Determining the parameter value after the next iteration based on the parameter value after the previous iteration and the weight decay term;

[0009] In response to determining that the parameter value after one iteration satisfies a preset condition, determining a corresponding target weight and target bias;

[0010] Determining an optimized convolutional neural network model according to the target weight and the target bias;

[0011] The concentration data of dissolved gas in transformer oil is obtained and processed based on the optimized convolutional neural network model to identify the type of transformer fault.

[0012] According to another aspect of the present application, a transformer fault identification device is provided, comprising:

[0013] A weight decay term determination module is configured to determine a weight decay term according to a parameter value after a previous iteration during an optimization process of the convolutional neural network model based on the optimization algorithm;

[0014] A module for determining a parameter value after the next iteration, configured to determine a parameter value after the next iteration based on the parameter value after the previous iteration and the weight decay term;

[0015] A target weight and target bias determination module is configured to determine the corresponding target weight and target bias when the parameter values ​​meet the preset conditions;

[0016] a convolutional neural network model optimization module, configured to determine an optimized convolutional neural network model based on the target weight and the target bias;

[0017] The fault type identification module is configured to obtain concentration data of dissolved gas in transformer oil and process the concentration data based on the optimized convolutional neural network model to identify the transformer fault type.

[0018] According to another aspect of the present application, an electronic device is provided, including:

[0019] at least one processor; and

[0020] a memory communicatively connected to the at least one processor; wherein,

[0021] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the transformer fault identification method described in any embodiment of the present application.

[0022] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the transformer fault identification method described in any embodiment of the present application when executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] FIG1 is a flow chart of a transformer fault identification method provided according to an embodiment of the present application;

[0024] FIG2 is a flow chart of another transformer fault identification method provided according to an embodiment of the present application;

[0025] FIG3 is a schematic structural diagram of a transformer fault identification device provided according to an embodiment of the present application;

[0026] FIG4 is a schematic structural diagram of an electronic device for implementing a transformer fault identification method according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] It should be noted that the terms "first", "second", "target", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] FIG1 is a flowchart of a transformer fault identification method according to an embodiment of the present application. The embodiment of the present application is applicable to the case of identifying the fault type of an oil-immersed transformer. The method can be executed by a transformer fault identification device, which can be implemented in the form of hardware and / or software. The transformer fault identification device can be configured in an electronic device with data processing capabilities. As shown in FIG1 , the method includes:

[0029] S110, in the process of optimizing the convolutional neural network model based on the optimization algorithm, determining a weight decay term according to the parameter value after the last iteration.

[0030] The optimization algorithm may be an Adaptive Moment Estimation (Adam) optimization algorithm, which is used to determine parameters of a convolutional neural network model, such as weights and biases.

[0031] The parameter value may be the parameter value θ in the Adam optimization algorithm. For example, the parameter value may be the output result of the model when the model is trained. For example, a parameter value of 1 indicates no fault, and a parameter value of 2 indicates a high temperature fault.

[0032] The weight decay term may be a polynomial used to increase the iteration speed of the parameter value when calculating the parameter value.

[0033] In the embodiment of the present application, it should be noted that the optimization process of the model corresponds to the iterative process of the parameter value. In the iterative process of the parameter value, the parameter value of the next iteration is equal to the difference between the parameter value of the previous iteration and other polynomials, and the other polynomials include weight decay terms.

[0034] For example, appropriate mathematical operations may be performed on the parameter values ​​after the last iteration to obtain a weight attenuation term, and the specific operation method may be multiplication by an associated attenuation coefficient, etc.

[0035] S120: Determine the parameter value after the next iteration according to the parameter value after the previous iteration and the weight decay term.

[0036] For example, in the past, during the iterative calculation of parameter values, the parameter value after the next iteration was determined based on the parameter value after the previous iteration. In the embodiment of the present application, the parameter value after the next iteration is determined based on the parameter value after the previous iteration and the weight decay term. The technical solution of the embodiment of the present application adds a process of subtracting the weight decay term when calculating the parameter value after the next iteration, thereby speeding up the iteration of the parameter value.

[0037] S130: When the parameter value satisfies a preset condition, a corresponding target weight and target bias are determined.

[0038] The preset conditions are conditions that reflect the completion of model optimization and can be determined based on actual conditions. For example, during the parameter value iteration process, if a preset number of iterations has been reached and convergence has occurred, it can be determined that the parameter value meets the preset conditions, and the corresponding target weight and target bias can be determined in this case.

[0039] In another feasible solution, as the parameter value iterates, it approaches the label corresponding to the sample. Based on the difference between the parameter value and the label, it can be determined whether the parameter value meets the preset condition. For example, if the label is determined to be 1, it indicates no fault. For this training sample, during the parameter value iteration process, if the parameter value is determined to be 2, it can be determined that the parameter value does not meet the preset condition. If the parameter value is determined to be 1.01, it can be determined that the parameter value meets the preset condition.

[0040] S140: Determine an optimized convolutional neural network model based on the target weight and the target bias.

[0041] For example, after obtaining the target weight and the target bias, the optimized convolutional neural network model is determined according to the target weight and the target bias.

[0042] S150, obtaining concentration data of dissolved gas in transformer oil, and processing the concentration data based on the optimized convolutional neural network model to identify the type of transformer fault.

[0043] For example, the concentration data of dissolved gas in transformer oil is obtained, and the concentration data is processed based on the optimized convolutional neural network model to obtain the transformer fault type.

[0044] The transformer may be an oil-immersed transformer. After various types of faults occur in such a transformer, the concentration of dissolved gas in the transformer oil may also change. Therefore, the fault type of the transformer may be determined by the concentration data of dissolved gas in the transformer oil.

[0045] Dissolved gas in oil refers to the gas dissolved in the oil in the transformer in a molecular state. The sources of dissolved gas in oil are mainly the dissolution of air, gas production due to normal operation and gas production due to internal faults. The dissolved gas components in transformer oil mainly include N2, O2, H2, CH4, C2H2, C2H4, C2H6, C3H6, C3H8, CO, CO2, etc. For example, the concentration data of dissolved gas in transformer oil obtained in the embodiment of the present application can be the concentration data of H2, CH4, C2H6, C2H4 and C2H2.

[0046] For example, after detecting that the fault identification control has been triggered, an input box for the concentration of dissolved gas in oil can be displayed to obtain the concentration data of dissolved gas in transformer oil input by the staff through the input box. In another feasible solution, an input interface for the concentration of dissolved gas in oil is provided, and the input interface is connected to a concentration acquisition device for dissolved gas in oil, so that the concentration data of dissolved gas in oil can be read in real time. Of the above two solutions, one can obtain the concentration data of dissolved gas in transformer oil input by the staff. The concentration data can be relatively accurate concentration data after calibration, and the type of transformer fault can be accurately identified through subsequent steps; the other can realize the real-time acquisition of the concentration data of dissolved gas in transformer oil, and thus can detect in real time whether the transformer has a fault.

[0047] For example, data on the concentration of dissolved gas in transformer oil is input into an optimized convolutional neural network model. The optimized convolutional neural network model processes the concentration data and generates an output result, which can be an output code. Different output codes correspond to different fault types. Exemplarily, the fault type includes at least one of the following: no fault, medium-low temperature overheating between 150°C and 700°C, high temperature overheating greater than 700°C, low-energy discharge, high-energy discharge, and partial discharge.

[0048] The technical solution of the embodiment of the present application includes: in the process of optimizing a convolutional neural network model based on an optimization algorithm, determining a weight decay term based on the parameter value after the previous iteration; determining the parameter value after the next iteration based on the parameter value after the previous iteration and the weight decay term; determining the corresponding target weight and target bias when the parameter value meets the preset conditions; determining the optimized convolutional neural network model based on the target weight and the target bias; obtaining the concentration data of the dissolved gas in the transformer oil, and processing the concentration data based on the optimized convolutional neural network model to identify the type of transformer fault. By adding the weight decay term when iteratively calculating the parameter value, this technical solution improves the iteration speed of the parameter value, thereby reducing the optimization time of the model, enabling the optimization algorithm to converge quickly, and improving the recognition accuracy of the model.

[0049] FIG2 is a flow chart of another transformer fault identification method provided in an embodiment of the present application. The embodiment of the present application is refined based on the above embodiment.

[0050] As shown in FIG2 , the method of the embodiment of the present application includes the following steps:

[0051] S210: The product of the parameter value after the previous iteration, the weight decay coefficient, and the learning rate is used as the weight decay term.

[0052] For example, the weight decay term can be expressed as: λlrθ t-1 ;

[0053] Among them, λ is the weight decay coefficient, lr is the learning rate, θ t-1 is the parameter value after the last iteration.

[0054] S220 , subtract the weight decay term from the parameter value after the previous iteration, and then subtract the target polynomial to obtain the parameter value after the next iteration.

[0055] The target polynomial is:

[0056] Among them, lr is the learning rate, M t is the partial first moment estimate of the gradient, V t is the partial second moment estimate of the gradient, and ε is the minimum value.

[0057] For example, the parameter value θ after the next iteration t It can be expressed by the following formula:

[0058] Among them, θ t is the parameter value of the tth iteration; θ t-1 is the parameter value of the t-1th iteration; lr is the learning rate, M tis the partial first moment estimate of the gradient, V t is the partial second moment estimate of the gradient, ∈ is the minimum value, and λ is the weight attenuation coefficient.

[0059] For example, the partial first-order moment estimate and the partial second-order moment estimate of the gradient are the first-order moment estimate and the second-order moment estimate of the amplified gradient, respectively.

[0060] For example, since the first-order moment estimate and the second-order moment estimate of the gradient will cause model loss during the iteration process, that is, deviate from the true value, the first-order moment estimate and the second-order moment estimate of the gradient are adaptively amplified to obtain the partial first-order moment estimate and the partial second-order moment estimate of the gradient, and then subsequent operations are performed to reduce the loss of the model.

[0061] For example, the process of determining the partial first-order moment estimate of the gradient includes: determining a first parameter based on the difference between a preset constant and an attenuation coefficient corresponding to the first-order moment estimate; dividing the first-order moment estimate of the gradient by the first parameter to obtain the partial first-order moment estimate of the gradient.

[0062] The preset constant can be determined according to actual conditions. For example, the preset constant is 1.

[0063] The partial first moment estimate of the gradient is calculated according to the following formula:

[0064] Among them, M t is the partial first moment estimate of the gradient, m t is the first moment estimate of the gradient, is the attenuation coefficient corresponding to the first-order moment estimate.

[0065] For example, the process of determining the partial second moment estimate of the gradient includes: determining a second parameter based on the difference between a preset constant and an attenuation coefficient corresponding to the second moment estimate; dividing the second moment estimate of the gradient by the second parameter to obtain the partial second moment estimate of the gradient.

[0066] The preset constant can be determined according to actual conditions. For example, the preset constant is 1.

[0067] The partial second moment estimate of the gradient is calculated according to the following formula:

[0068] Among them, V t is the partial second moment estimate of the gradient, v t is the second moment estimate of the gradient, is the attenuation coefficient corresponding to the second-order moment estimate.

[0069] S230: When the parameter value satisfies the preset conditions, the corresponding target weight and target bias are determined.

[0070] S240: Determine an optimized convolutional neural network model based on the target weight and the target bias.

[0071] S250: Acquire concentration data of dissolved gas in transformer oil, and process the concentration data based on the optimized convolutional neural network model to identify the type of transformer fault.

[0072] For example, after obtaining the concentration data of dissolved gas in transformer oil, the method also includes: converting the integer digits and decimal digits of the dissolved gas concentration data into a binary form of a first preset number of bits; generating a two-dimensional array based on the integer digits and decimal digits of the binary form of the dissolved gas concentration data; accordingly, processing the concentration data based on the optimized convolutional neural network model, including: processing the two-dimensional array based on the optimized convolutional neural network model.

[0073] For example, after obtaining the concentration data of dissolved gases in transformer oil, the concentration data is converted into binary, and the number of integer digits and decimal digits are both the first preset number of digits. For example, the first preset number of digits is 10, so 10 integer digits and 10 decimal digits are obtained; then, according to the binary concentration data of the five dissolved gases, a two-dimensional array is generated, that is, a 10*10 two-dimensional array, and then the two-dimensional array is processed based on the optimized convolutional neural network model to obtain the fault type.

[0074] The technical solution of the embodiment of the present application converts the dissolved gas concentration data into binary form and then generates a two-dimensional array, thereby converting a one-dimensional data set into a two-dimensional image, solving the problem of poor training effect with small samples; and also achieving the effect of data enhancement.

[0075] Figure 3 is a schematic diagram of the structure of a transformer fault identification device provided in one embodiment of the present application. The device can execute the transformer fault identification method provided in any embodiment of the present application and has the corresponding functional modules and beneficial effects of the execution method. As shown in Figure 3, the device includes:

[0076] A weight decay term determination module 310 is configured to determine a weight decay term based on a parameter value after a previous iteration during the optimization process of the convolutional neural network model based on the optimization algorithm;

[0077] The parameter value determination module 320 after the next iteration is configured to determine the parameter value after the next iteration based on the parameter value after the previous iteration and the weight decay term;

[0078] The target weight and target bias determination module 330 is configured to determine the corresponding target weight and target bias when the parameter value meets the preset conditions;

[0079] The convolutional neural network model optimization module 340 is configured to determine an optimized convolutional neural network model based on the target weight and the target bias.

[0080] The fault type identification module 350 is configured to obtain concentration data of dissolved gas in transformer oil and process the concentration data based on the optimized convolutional neural network model to identify the transformer fault type.

[0081] For example, the weight decay term determination module 310 includes:

[0082] The weight decay term determination unit is configured to use the product of the parameter value after the previous iteration, the weight decay coefficient, and the learning rate as the weight decay term.

[0083] For example, the parameter value determination module 320 after the next iteration includes:

[0084] The parameter value determination unit after the next iteration is configured to obtain the parameter value after the next iteration by subtracting the weight decay term from the parameter value after the previous iteration and then subtracting the target polynomial; the target polynomial is:

[0085] Among them, lr is the learning rate, M t is the partial first moment estimate of the gradient, V t is the partial second moment estimate of the gradient, and ε is the minimum value.

[0086] For example, the partial first-order moment estimate and the partial second-order moment estimate of the gradient are the first-order moment estimate and the second-order moment estimate of the amplified gradient, respectively.

[0087] For example, the parameter value determination unit after the next iteration includes a determination subunit of the partial first-order moment estimation of the gradient, which is set to:

[0088] Determining a first parameter based on a difference between a preset constant and an attenuation coefficient corresponding to a first-order moment estimate;

[0089] The first-order moment estimate of the gradient is divided by the first parameter to obtain a partial first-order moment estimate of the gradient.

[0090] For example, the parameter value determination unit after the next iteration includes a determination subunit of the partial second moment estimation of the gradient, which is set to:

[0091] determining a second parameter based on a difference between a preset constant and an attenuation coefficient corresponding to the second-order moment estimate;

[0092] The second-order moment estimate of the gradient is divided by the second parameter to obtain a partial second-order moment estimate of the gradient.

[0093] For example, the device further includes a binary form conversion module and a two-dimensional array generation module.

[0094] a binary form conversion module configured to convert integer digits and decimal digits of dissolved gas concentration data into a binary form of a first preset number of digits;

[0095] a two-dimensional array generating module configured to generate a two-dimensional array according to integer digits and decimal digits of concentration data in binary form of dissolved gas;

[0096] The fault type identification module 350 is configured to process the two-dimensional array based on the optimized convolutional neural network model.

[0097] A transformer fault identification device provided in an embodiment of the present application can execute a transformer fault identification method provided in any embodiment of the present application, and has functional modules and beneficial effects corresponding to the execution method.

[0098] FIG4 shows a block diagram of an electronic device 10 that can be used to implement an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0099] As shown in FIG4 , the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 and a random access memory (RAM) 13, that is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0100] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0101] The processor 11 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the transformer fault identification method.

[0102] In some embodiments, the transformer fault identification method can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the transformer fault identification method described above can be performed. For example, in other embodiments, processor 11 can be configured to execute the transformer fault identification method in any other suitable manner (e.g., via firmware).

[0103] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0104] Computer programs for implementing the methods of the present application may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0105] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. For example, a computer-readable storage medium can be a machine-readable signal medium. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. A computer-readable storage medium can be a non-transient computer-readable storage medium.

[0106] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0107] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0108] A computing system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. The client-server relationship is established by computer programs running on the respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within a cloud computing service system that addresses the management difficulties and poor business scalability of physical hosts and virtual private server (VPS) services in related technologies.

[0109] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the multiple steps described in this application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of this application can be achieved. This is not limited herein.

Claims

1. A transformer fault identification method, comprising: In the process of optimizing the convolutional neural network model based on the optimization algorithm, the weight decay term is determined according to the parameter value after the previous iteration; Determining the parameter value after the next iteration according to the parameter value after the previous iteration and the weight decay term; In response to determining that the parameter value after the next iteration satisfies a preset condition, determining a corresponding target weight and target bias; Determining an optimized convolutional neural network model according to the target weight and the target bias; Concentration data of dissolved gas in transformer oil is obtained, and the concentration data is processed based on the optimized convolutional neural network model to identify the type of transformer fault.

2. The method according to claim 1, wherein: Determining the weight decay term according to the parameter value after the last iteration includes: The product of the parameter value after the previous iteration, the weight decay coefficient and the learning rate is used as the weight decay term.

3. The method according to claim 1, wherein: The step of determining the parameter value after the next iteration according to the parameter value after the previous iteration and the weight attenuation term includes: The parameter value after the previous iteration is subtracted from the weight attenuation term, and then subtracted from the target polynomial to obtain the parameter value after the next iteration; the target polynomial is: Among them, lr is the learning rate, M t is the partial first-order moment estimate of the gradient, V t is the partial second moment estimate of the gradient, ∈ is the minimum value.

4. The method according to claim 3, wherein: The partial first-order moment estimate of the gradient is the first-order moment estimate of the amplified gradient, and the partial second-order moment estimate of the gradient is the second-order moment estimate of the amplified gradient.

5. The method according to claim 4, wherein: The process of determining the partial first-order moment estimate of the gradient includes: Determine a first parameter according to a difference between a preset constant and an attenuation coefficient estimated corresponding to a first-order moment of the gradient; The first-order moment estimate of the gradient is divided by the first parameter to obtain a partial first-order moment estimate of the gradient.

6. The method according to claim 4, wherein: The process of determining the partial second-order moment estimation of the gradient includes: Determine a second parameter according to a difference between a preset constant and an attenuation coefficient estimated corresponding to a second-order moment of the gradient; The second-order moment estimate of the gradient is divided by the second parameter to obtain a partial second-order moment estimate of the gradient.

7. The method according to claim 1, wherein: After obtaining the concentration data of dissolved gas in transformer oil, the method further includes: Converting the integer digits and decimal digits of the concentration data into a binary form of a first preset number of digits; Generate a two-dimensional array according to the integer digits and decimal digits of the concentration data in binary form; The processing of the concentration data based on the optimized convolutional neural network model includes: The two-dimensional array is processed based on the optimized convolutional neural network model.

8. A transformer fault identification device, comprising: A weight decay term determination module is configured to determine a weight decay term according to a parameter value after a previous iteration during an optimization process of a convolutional neural network model based on an optimization algorithm; A module for determining the parameter value after the next iteration, configured to determine the parameter value after the next iteration according to the parameter value after the previous iteration and the weight decay term; A target weight and target bias determination module, configured to determine a corresponding target weight and target bias in response to determining that the parameter value after the next iteration satisfies a preset condition; A convolutional neural network model optimization module, configured to determine an optimized convolutional neural network model according to the target weight and the target bias; The fault type identification module is configured to obtain concentration data of dissolved gas in transformer oil and process the concentration data based on the optimized convolutional neural network model to identify the transformer fault type.

9. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the transformer fault identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a processor to implement the transformer fault identification method according to any one of claims 1 to 7 when executed.

Images