Partial discharge waveform classification method and system based on convolution gating circulation network

Abstract

The invention discloses a partial discharge waveform classification method and system based on a convolution gated cycle network, and the method comprises the steps: receiving electromagnetic waves radiated by partial discharge when equipment has defects and partial discharge occurs; electromagnetic waves radiated by partial discharge are coupled into a time-domain waveform signal, and then the time-domain waveform signal passes through a one-dimensional convolutional network and a gated cycle network to realize defect identification of partial discharge. Characteristic extraction of the partial discharge waveform is realized, the accuracy of the defect identification method is improved, non-intrusive and high-anti-interference intelligent diagnosis of the discharge state is realized, and intelligent operation and maintenance of a power system are facilitated.

Description

Technical Field

[0001] The present invention belongs to the technical field of GIS insulation defect fault diagnosis, and in particular relates to a partial discharge waveform classification method and system based on a convolutional gated recurrent network. Background Art

[0002] In the past, gas-insulated metal-enclosed switchgear (GIS) was widely considered a type of power equipment requiring little or no maintenance. By enclosing critical equipment such as circuit breakers, disconnectors, earthing switches, and transformers and filling them with SF6 gas, it significantly improves insulation performance, leading to its widespread use in power transmission and distribution systems. However, in the complex operating environment of power systems, the actual failure frequency of GIS is far higher than expected. Various types of insulation aging and defects may exist within GIS, leading to partial discharges that seriously impact the safe operation of power systems.

[0003] The main forms of partial discharge (PD) within GIS include metal tip discharge, metal particle discharge, suspended potential discharge, gas cavity discharge, and insulator surface discharge. Because different types of PD occur in different locations and significantly vary in the extent of damage to the GIS, identifying the discharge type based on the electromagnetic wave signals generated by PD can help provide early warning of potential faults, assess risks, and avoid economic losses. This can guide the power system in taking appropriate measures and is of great significance to the safe operation and maintenance of the power system.

[0004] When partial discharge occurs in the power system, the transient electromagnetic waves radiated from it in the range of 300MHz to 3GHz are coupled and converted into electrical signals through the broadband antenna built into the UHF sensor. The sensor captures the waveform details at a high sampling rate and transmits them to the preprocessing unit for noise reduction, normalization, and segmented interception. The acquired waveform is then analyzed to determine the type of discharge defect.

[0005] Existing traditional partial discharge signal classification methods rely too heavily on the physical characteristics of electromagnetic wave signals, such as the main discharge frequency, discharge energy, pulse amplitude, and rise time. These methods are less effective at recognizing signals with less distinct characteristics, suffer from feature redundancy, poor generalization, and are unable to adapt to waveform variations under varying operating conditions. Newer machine learning methods have limited ability to model long-range temporal dependencies in discharge waveforms (such as the periodicity and intermittency of pulse trains), resulting in low accuracy in recognizing complex discharge patterns. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to address the deficiencies in the above-mentioned prior art and provide a method and system for partial discharge waveform classification based on a convolutional gated recurrent network, which is used to solve the technical problems of the existing partial discharge waveform classification methods, such as low accuracy in complex discharge pattern recognition and weak long-range time dependence.

[0007] The present invention adopts the following technical solutions:

[0008] The partial discharge waveform classification method based on convolutional gated recurrent network includes the following steps:

[0009] When a defect occurs in the equipment and partial discharge occurs, the device receives the electromagnetic waves radiated by the partial discharge;

[0010] The electromagnetic waves radiated by partial discharge are coupled into time domain waveform signals, and then passed through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect recognition.

[0011] Preferably, the one-dimensional convolutional network includes multiple one-dimensional convolutional layers and pooling layers. The one-dimensional convolutional layer extracts local waveform features by sliding convolution kernels, and the pooling layer downsamples the feature map and compresses the length of the feature map. By cooperating with the one-dimensional convolutional layer, the perception range of the input waveform is expanded.

[0012] After multiple one-dimensional convolutions and pooling, the one-dimensional convolutional network extracts the time domain waveform features, denoted as X, and uses the extracted time domain waveform features as the input of the gated recurrent network;

[0013] Each gate control unit receives a time domain waveform feature as input and obtains the output h through calculation; for the t-th gate control unit, it resets the gate r t The current gate input and the previous gate output h t-1 The linear transformation is added and activated by the sigmoid function;

[0014] For the tth gate control unit, it updates the gate z t By the current gate input x t and the output h of the previous gate t-1 The linear transformation is added and activated by the sigmoid function;

[0015] The gate control unit generates the candidate hidden state h of the current layer by comprehensively updating the information of the gate and the reset gate. t ;

[0016] By updating the gate to fuse the candidate state with the historical state, the output h of the gate control unit can be obtained t ;

[0017] For each gating sequence, take the output h of the last gating unit n As a characterization of waveform characteristics.

[0018] Preferably, the neurons in each layer receive signal input from the neurons in the previous layer, and output it to the neurons in the next layer after discrete convolution. The mathematical expression is as follows:

[0019]

[0020] Among them, V i k represents the output of the i-th neuron in the k-th layer, W is the one-dimensional convolution kernel, b is the bias, and σ(·) is the convolution activation function.

[0021] Preferably, the selected pooling method is maximum pooling, which retains the local maximum value by sliding the window. Assume that the pooling window size is j and the maximum pooling P j as follows:

[0022] P j =max(V i,s ,V i,s+1 ,...,V i,s+j )

[0023] Among them, V i,s ,V i,s+1 ,...,V i,s+j for each element in the pooling window.

[0024] Preferably, the reset gate r of the tth gating unit t By the current gate input X t and the output h of the previous gate t-1 The linear transformation is added and the sigmoid function is used to activate the reset gate r t :

[0025] r t =sigmoid(W reset X t +U reset h t-1 +b reset )

[0026] Among them, W reset , U reset is the weight matrix of the reset gate, b reset is the offset.

[0027] Preferably, the update gate z of the tth gating unit t By the current gate input x t and the output h of the previous gate t-1 The linear transformation is added and activated with the sigmoid function:

[0028] z t =sigmoid(W update X t +U update h t-1 +b update )

[0029] Among them, W update , U update is the weight matrix of the update gate, b update is the offset.

[0030] Preferably, the candidate hidden state of the t-th gate unit is for:

[0031]

[0032] Among them, W hidden , U hidden is the weight matrix of the candidate hidden state, b hidden is the bias, and ⊙ is the element-by-element multiplication.

[0033] Preferably, the output ht of the t-th gating unit is:

[0034]

[0035] Among them, z t is the update gate of the t-th gate control unit, h t-1 is the output of the t-1th gate unit, ⊙ is the element-by-element multiplication, is the candidate hidden state of the t-th gated unit.

[0036] Preferably, the cross entropy loss function is used to train the one-dimensional convolutional network and the gated recurrent network, and the loss function is as follows:

[0037]

[0038] in, is the symbolic function, P ic The network calculates the probability that the i-th sample belongs to the c-th discharge type during training, M is the total number of discharge types, L i is the loss value of the i-th sample, and N is the total number of samples.

[0039] In a second aspect, an embodiment of the present invention provides a partial discharge waveform classification system based on a convolutional gated recurrent network, comprising:

[0040] The acquisition module receives the electromagnetic waves radiated by partial discharge when a defect occurs in the equipment;

[0041] The recognition module couples the electromagnetic waves radiated by partial discharge into time domain waveform signals, and then passes them through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect recognition.

[0042] In a third aspect, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned method for partial discharge waveform classification based on a convolutional gated recurrent network are implemented.

[0043] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium comprising a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for partial discharge waveform classification based on a convolutional gated recurrent network.

[0044] In a fifth aspect, a chip comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned method for classifying partial discharge waveforms based on a convolutional gated recurrent network are implemented.

[0045] In a sixth aspect, an embodiment of the present invention provides an electronic device, comprising a computer program, which, when executed by the electronic device, implements the steps of the above-mentioned method for partial discharge waveform classification based on a convolutional gated recurrent network.

[0046] Compared with the prior art, the present invention has at least the following beneficial effects:

[0047] A partial discharge waveform classification method based on a convolutional gated recurrent network (CRN) is proposed. One-dimensional convolution excels at extracting local details in waveforms, while gated recurrent networks can capture temporal dependencies. The combination of the two focuses on both details and overall trends, improving classification accuracy. For non-stationary, noisy waveforms received by sensors, the model automatically identifies high-frequency and low-frequency features using multi-scale convolution kernels and memorizes the long- and short-term correlations of effective signals through a CRN, filtering out environmental noise interference and enhancing robustness. Features are learned directly from the original waveform, avoiding the limitations of traditional methods that rely on manually designed statistical parameters, lowering the threshold for expert experience and improving analysis efficiency. The model has a streamlined structure and low computational complexity, making it suitable for deployment in embedded devices or edge terminals to enable online, real-time monitoring and early warning of partial discharge waveforms, meeting the timeliness requirements of power equipment status monitoring. Through the innovative combination of "local feature + temporal modeling," the method addresses the problems of traditional technologies' insufficient ability to recognize complex waveforms, poor anti-interference performance, and reliance on manual labor, providing a high-precision, highly robust intelligent diagnosis solution for power equipment defects. The CRN is particularly suitable for online monitoring systems in complex electromagnetic environments, facilitating safe power grid operation and fault prevention.

[0048] Furthermore, the one-dimensional convolution layer accurately captures local details in the waveform through sliding convolution kernels, and the pooling layer gradually compresses the length of the feature map, which not only retains key information but also expands the perception range and enhances the robustness to the overall trend of the waveform; the pooling operation significantly reduces the feature dimension, reduces the sequence length of subsequent gated recurrent network processing, and improves the overall computational efficiency; through the hierarchical local feature extraction of one-dimensional convolution and the dynamic gated timing fusion of the gated recurrent network, high-precision and high-efficiency waveform classification is achieved, which has noise resistance, lightweight and strong timing modeling capabilities, and provides reliable technical support for online monitoring and defect diagnosis of partial discharge of power equipment.

[0049] Furthermore, the local feature extraction capability of the one-dimensional convolution module and the long-term time series modeling capability of the gating unit are deeply integrated. In the training of the same local discharge waveform samples, the model outperforms other similar models. The one-dimensional convolution module uses multi-scale convolution kernels to capture local details such as the pulse front and oscillation frequency of the discharge waveform. The gated loop module uses the gating mechanism to analyze the time correlation of the discharge pulse group and realizes joint optimization through feature cascade. Compared with the long short-term memory network, the use of the gated network reduces the number of parameters and computational complexity, which is conducive to deployment on power equipment and provides a high-precision, low-latency, and highly robust solution for intelligent diagnosis of power equipment.

[0050] It can be understood that the beneficial effects of the second to sixth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0051] In summary, the present invention realizes the feature extraction of partial discharge waveforms, improves the accuracy of the defect identification method, and implements non-invasive, highly anti-interference intelligent diagnosis of discharge status, which is beneficial to the intelligent operation and maintenance of power systems.

[0052] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0054] Figure 1 Flow chart of the method of the present invention;

[0055] Figure 2 The following is a 10-fold cross-validation result diagram of the accuracy of partial discharge recognition of the five models;

[0056] Figure 3Box plots of the accuracy of identifying partial discharge for the five models;

[0057] Figure 4 A schematic diagram of a computer device provided in accordance with an embodiment of the present invention;

[0058] Figure 5 The present invention is a block diagram of an electronic device according to an embodiment of the present invention.

[0059] Among them, 60. Computer device; 61. Processor; 62. Memory; 63. Computer program; 600. Electronic device; 610. Processing unit; 620. Storage unit; 6201. Random access memory unit; 6202. Cache memory unit; 6203. Read-only memory unit; 6204. Program / Utility; 6205. Program module; 630. Bus; 640. Display unit; 650. Input / output interface; 660. Network adapter; 700. External device. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] In the description of the present invention, it is to be understood that the terms “include” and “comprise” indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0062] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0063] It should be further understood that the term "and / or" as used in the present specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects are in an "or" relationship.

[0064] It should be understood that although the terms "first," "second," and "third" may be used to describe preset ranges in embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are merely used to distinguish one preset range from another. For example, without departing from the scope of embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0065] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0066] The accompanying drawings illustrate various schematic diagrams of structures according to embodiments disclosed herein. These figures are not drawn to scale; for clarity, some details are exaggerated and some details may be omitted. The shapes of the various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art may design regions / layers with different shapes, sizes, and relative positions as needed.

[0067] The present invention provides a partial discharge waveform classification method based on a convolutional gated recurrent network. The method uses power equipment connected to a UHF sensor. When a defect occurs in the equipment and partial discharge occurs, the sensor receives the electromagnetic waves radiated by the partial discharge, which are coupled into a time domain waveform signal through the sensor's built-in broadband antenna. The signal then passes through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect identification.

[0068] Example 1

[0069] See also Figure 1 The present invention provides a partial discharge waveform classification method based on a convolutional gated recurrent network, comprising the following steps:

[0070] S1. When a defect occurs in the equipment and partial discharge occurs, the sensor receives the electromagnetic waves radiated by the partial discharge;

[0071] S2. The sensor's built-in broadband antenna couples the signal into a time-domain waveform, which is then transformed into a one-dimensional convolutional network and a gated recurrent network to identify partial discharge defects.

[0072] S201, the one-dimensional convolutional network includes multiple one-dimensional convolutional layers and pooling layers. The one-dimensional convolutional layer extracts local waveform features through sliding convolution kernels. The neurons in each layer receive signal input from the neurons in the previous layer, and after discrete convolution, output it to the neurons in the next layer. The pooling layer downsamples the feature map and compresses the length of the feature map. By cooperating with the one-dimensional convolutional layer, the model gradually expands the perception range of the input waveform.

[0073] Each layer of neurons receives signal input from the neurons in the previous layer, and outputs it to the neurons in the next layer after discrete convolution. Its mathematical expression is as follows:

[0074]

[0075] Among them, V i k represents the output of the i-th neuron in the k-th layer, W is the one-dimensional convolution kernel, b is the bias, and σ(·) is the convolution activation function.

[0076] Preferably, the activation function is set to the relu function, and the expression is as follows:

[0077]

[0078] Preferably, the selected pooling method is maximum pooling, which retains the local maximum by sliding the window. Assuming the pooling window size is j, the maximum pooling expression is as follows:

[0079] P j =max(V i,s ,V i,s+1 ,...,V i,s+j ) (3)

[0080] Among them, V i,s ,V i,s+1 ,...,V i,s+j for each element in the pooling window.

[0081] S202. After multiple one-dimensional convolutions and pooling operations, the one-dimensional convolutional network extracts a time-domain waveform feature, which is denoted as X. The extracted time-domain waveform feature is used as the input of the gated recurrent network.

[0082] The gated recurrent network includes several gated recurrent units, and the gated recurrent unit includes a reset gate and an update gate.

[0083] S203, each gate control unit receives a time domain waveform feature as input and obtains an output h through calculation; for the t-th gate control unit, it resets the gate r t The current gate input and the previous gate output h t-1 The linear transformation is added and activated by the sigmoid function;

[0084] Use the sigmoid function to activate the reset gate r t :

[0085] r t =sigmoid(W reset X t +U reset h t-1 +b reset ) (4)

[0086] Among them, W reset , U reset is the weight matrix of the reset gate, b reset is the offset.

[0087] The Sigmoid function expression is as follows:

[0088]

[0089] The Sigmoid function maps the input to the range of 0 to 1. At this time, the value represents the proportion of information to be reset, which is used to control the contribution of the historical hidden state to the current candidate state. t When it approaches 0, it means that the historical state will be largely ignored and the focus will be on the current input.

[0090] On the contrary, when the reset gate r t When it approaches 1, it indicates that the historical state is retained, and the calculation result of the unit will take into account both the current input and the historical input information.

[0091] S204: For the t-th gate control unit, it updates the gate z t By the current gate input x t and the output h of the previous gate t-1 The linear transformation is added and activated by the sigmoid function;

[0092] z t =sigmoid(W update X t +U update h t-1 +b update ) (6)

[0093] Among them, W update , U update is the weight matrix of the update gate, b update is the offset.

[0094] S205, the gate control unit generates the candidate hidden state of the current layer by comprehensively updating the information of the gate and the reset gate

[0095]

[0096] Among them, W hidden , U hidden is the weight matrix of the candidate hidden state, b hidden is the bias, and ⊙ represents element-by-element multiplication.

[0097] The candidate hidden state controls the influence of the historical state on the current state through the reset gate. In discharge waveform classification, if a certain segment is irrelevant in timing (such as a noise segment), the state is closed through the reset gate to prevent the propagation of erroneous information.

[0098] S206: By updating the gate to fuse the candidate state with the historical state, the output h of the gate control unit can be obtained. t ;

[0099]

[0100] Among them, z t is the update gate of the t-th gate control unit, h t-1 is the output of the t-1th gate unit, ⊙ is the element-by-element multiplication, is the candidate hidden state of the t-th gated unit.

[0101] S207. For each gating sequence, take the output h of the last gating unit n As a characterization of waveform characteristics.

[0102] Since the one-dimensional convolution process broadens the number of waveform channels, each waveform obtains more waveform features. Therefore, after passing through the gated recurrent network, each waveform contains more gated outputs. The gated outputs are reduced to 5 neurons through the multi-layer perceptron, representing 5 typical types of partial discharge defects.

[0103] This model is suitable for partial discharge defect classification of gas-insulated metal-enclosed switchgear. Preferably, a cross entropy loss function is used for network training. The loss function is as follows:

[0104]

[0105] in, is a sign function. If the i-th sample belongs to the c-th discharge type, the value is 1, otherwise it is 0. ic The probability that the i-th sample belongs to the c-th discharge type is calculated for the network during training. M is the total number of discharge types. For example, when using the model to classify 5 typical partial discharges, M is 5. i is the loss value of the i-th sample, and N is the total number of samples.

[0106] Those skilled in the art will appreciate that various aspects of the present invention may be implemented as systems, methods, or program products. Accordingly, various aspects of the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, microcode, etc.), or in a combination of hardware and software, collectively referred to herein as "circuits," "modules," or "platforms."

[0107] Example 2

[0108] The present invention provides a partial discharge waveform classification system based on a convolutional gated recurrent network. The system can be used to implement the above-mentioned partial discharge waveform classification method based on a convolutional gated recurrent network. Specifically, the partial discharge waveform classification system based on a convolutional gated recurrent network includes an acquisition module and an identification module.

[0109] Among them, the acquisition module receives the electromagnetic waves radiated by the partial discharge when a defect occurs in the equipment;

[0110] The recognition module couples the electromagnetic waves radiated by partial discharge into time domain waveform signals, and then passes them through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect recognition.

[0111] Example 3

[0112] The present invention provides a terminal device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be other general-purpose processors, graphics processing units (GPUs), tensor processing units (TPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to implement corresponding method processes or corresponding functions; the processor described in the embodiment of the present invention can be used for the operation of a partial discharge waveform classification method based on a convolutional gated recurrent network, including:

[0113] When a defect occurs in the equipment and partial discharge occurs, the electromagnetic waves radiated by the partial discharge are received; the electromagnetic waves radiated by the partial discharge are coupled into a time domain waveform signal, and then passed through a one-dimensional convolutional network and a gated recurrent network to realize the defect identification of partial discharge.

[0114] See also Figure 4 The terminal device is a computer device. The computer device 60 of this embodiment includes: a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable by the processor 61. When executed by the processor 61, the computer program 63 implements the partial discharge waveform classification method based on a convolutional gated recurrent network in the embodiment. To avoid repetition, a detailed description is not given here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the partial discharge waveform classification system based on a convolutional gated recurrent network in the embodiment. To avoid repetition, a detailed description is not given here.

[0115] The computer device 60 may be a desktop computer, a notebook computer, a PDA, a cloud server, or other computing devices. The computer device 60 may include, but is not limited to, a processor 61 and a memory 62. It will be understood by those skilled in the art that Figure 4 This is merely an example of the computer device 60 and does not constitute a limitation of the computer device 60 . The computer device 60 may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0116] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, a graphics processing unit (GPU), a tensor processing unit (TPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0117] The memory 62 may be an internal storage unit of the computer device 60, such as a hard disk or memory of the computer device 60. The memory 62 may also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 60.

[0118] Furthermore, the memory 62 may include both an internal storage unit of the computer device 60 and an external storage device. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 may also be used to temporarily store data that has been output or is about to be output.

[0119] See also Figure 5 The terminal device is an electronic device 600, which is implemented as a general-purpose computing device. The components of the electronic device may include, but are not limited to, at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), and a display unit 640.

[0120] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present invention described in the above method section of this specification. For example, the processing unit 610 can perform the following steps: Figure 1 Follow the steps shown in .

[0121] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0122] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0123] Bus 630 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0124] The electronic device 600 may also communicate with one or more external devices 700 (e.g., a keyboard, a pointing device, a Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 600, and / or any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., a router, a modem). Such communication may occur via an input / output interface 650. Furthermore, the electronic device 600 may also communicate with one or more networks (e.g., a local area network, a wide area network, and / or a public network, such as the Internet) via a network adapter 660. The network adapter 660 may communicate with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0125] Example 4

[0126] The present invention also provides a storage medium, specifically a computer-readable storage medium, which is a memory device in a terminal device for storing programs and data. It is understood that the computer-readable storage medium herein may include both built-in storage media in the terminal device and, of course, extended storage media supported by the terminal device. It may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or component. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for being loaded and executed by a processor. These instructions may be one or more computer programs (including program code). It should be noted that more specific examples of the computer-readable storage medium herein include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0127] Computer-readable storage media also include data signals propagated in baseband or as part of a carrier wave, which carry readable program code. Such propagated data signals can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium can also be any readable medium other than a readable storage medium, which can send, propagate, or transmit programs for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the readable storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, radio frequency, etc., or any suitable combination of the above.

[0128] The program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, and the like, as well as conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device via any type of network, including a local area network or a wide area network, or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0129] The processor may load and execute one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the partial discharge waveform classification method based on a convolutional gated recurrent network in the above embodiment. The processor may load and execute the following steps:

[0130] When a defect occurs in the equipment and partial discharge occurs, the electromagnetic waves radiated by the partial discharge are received; the electromagnetic waves radiated by the partial discharge are coupled into a time domain waveform signal, and then passed through a one-dimensional convolutional network and a gated recurrent network to realize the defect identification of partial discharge.

[0131] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0132] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0133] The present invention provides a partial discharge waveform classification method based on a convolutional gated recurrent network as follows:

[0134] 1) Using ultra-high frequency sensors to collect partial discharge waveform signals of power equipment (such as GIS, cables, etc.) under various defects;

[0135] 2) Define discharge categories (such as corona discharge, surface discharge, air gap discharge, and suspension discharge) based on the physical mechanism of discharge, and have power equipment diagnostic experts annotate the original waveforms;

[0136] 3) De-noise and normalize the original waveform and classify it according to the annotated labels;

[0137] 4) Setting hyperparameters for one-dimensional convolutional networks and gated recurrent networks, such as learning rate, optimizer, and loss function;

[0138] 5) Divide the classified original waveforms into training and test sets, and train the one-dimensional convolutional network and gated recurrent network;

[0139] 6) Based on the accuracy of the test set, select the optimal one-dimensional convolution weight parameters and gated network parameters after training, and evaluate the performance of the trained one-dimensional convolutional network and gated recurrent network models through methods such as 10-fold cross-validation;

[0140] 7) Install a UHF sensor on the power equipment to be tested to collect and record partial discharge signals;

[0141] 8) Apply the same noise reduction and normalization methods as in training step 3) to the live waveform, and input the waveform into the trained one-dimensional convolutional gated training network;

[0142] 9) The network automatically calculates the probability distribution of each discharge type;

[0143] 10) Repeat the process from 7) to 9) to calculate the probability of discharge type for the partial discharge signals collected multiple times, confirm the possible discharge type, and realize fault diagnosis and early warning of power equipment.

[0144] To verify the performance of the proposed one-dimensional gated recurrent network, the network was trained using typical GIS partial discharge waveform data. A one-dimensional convolutional network, a recurrent neural network, a long short-term memory network, and a support vector machine were selected as comparison models. The partial discharge waveforms of five typical defects in gas-insulated metal-enclosed switchgear were classified. The training set was divided into a training and test set ratio of 7:3.

[0145] Table 1 The accuracy of different network training

[0146]

[0147] Table 1 shows the accuracy of the five networks after training. Among them, the one-dimensional gated recurrent network proposed in this paper has a higher accuracy rate than other networks, reaching 97.826%. This network extracts local pulse features through a one-dimensional shallow convolution kernel and combines a gated unit to capture long-range temporal context. Through cascade fusion, it realizes deep feature extraction of time-domain local discharge waveforms.

[0148] In order to avoid the accuracy error caused by the division of training set and validation set, the present invention carried out a 10-fold cross-validation method. This method divides the partial discharge waveform data set into 10 equal parts, selects one of them as the test set each time, and the remaining nine as the training set, and repeats 10 times. This method can ensure that each data set participates in training and validation, avoiding the deviation that may be caused by a single data division. The validation results of the five groups of data are shown in Figure 2. Figure 2 At the same time, multiple random trainings were conducted on each model, and random training sets and validation sets were divided. The box plots of the accuracy of multiple experiments are shown in Figure 3 shown.

[0149] Figure 2 and Figure 3The performance of five models in the task of GIS partial discharge waveform recognition is described. It can be seen that the accuracy of the one-dimensional gated recurrent network is generally higher than that of other networks, and the dispersion of the accuracy is small, indicating that the network can maintain accuracy while taking into account stability. The advantage of the method proposed in the present invention is that the waveform is first extracted through a one-dimensional convolution kernel to extract local details of the pulse waveform, such as the rising edge slope, oscillation decay rate, etc., retaining high-frequency features, and then the gated cycle is used to establish the long-range time dependence of the discharge pulse group, such as the periodic discharge information, the interval regularity of different types of discharges, etc. In this process, the reset gate is trained to automatically ignore invalid historical states during low signal-to-noise ratio periods to prevent noise propagation. At the same time, compared with the long short-term memory network, the number of parameters of the gated recurrent unit and the amount of computation of the model are significantly reduced, making the model convenient for on-site lightweight equipment deployment while ensuring recognition accuracy.

[0150] Compared with the model proposed in this paper, other models generally have lower accuracy. For example, one-dimensional convolutional networks only focus on the local discharge field of view and cannot effectively establish global temporal patterns. Recurrent neural networks lack a filtering mechanism for random noise and suffer from the problem of vanishing gradients during training. Support vector machines have difficulty adapting to nonlinear patterns in complex waveforms, resulting in limited feature expression capabilities.

[0151] In summary, the present invention's method and system for partial discharge waveform classification based on a convolutional gated recurrent network (CDGRN) leverages the collaborative design of 1D-CNN and GRU to balance local feature extraction with global temporal modeling. This method addresses the shortcomings of traditional technologies, such as insufficient complex waveform recognition, poor interference immunity, and low real-time performance. Its lightweight, end-to-end nature provides an efficient and reliable solution for online monitoring and intelligent diagnosis of partial discharge in power equipment, significantly enhancing the automation level and fault prevention capabilities of power grid operations and maintenance.

[0152] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0153] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0154] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0155] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical functional division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, and can be electrical, mechanical, or other forms.

[0156] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0157] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0158] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above-mentioned various method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0159] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0160] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0162] The above content is only for explaining the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A partial discharge waveform classification method based on a convolutional gated recurrent network, characterized in that: The following steps are involved: When a defect occurs in the equipment and partial discharge occurs, the device receives the electromagnetic waves radiated by the partial discharge; The electromagnetic waves radiated by partial discharge are coupled into time domain waveform signals, and then passed through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect recognition.

2. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 1, characterized in that: The one-dimensional convolutional network consists of multiple one-dimensional convolutional layers and pooling layers. The one-dimensional convolutional layer extracts local waveform features by sliding the convolution kernel. The pooling layer downsamples the feature map and compresses the length of the feature map. By cooperating with the one-dimensional convolutional layer, it expands the perception range of the input waveform. After multiple one-dimensional convolutions and pooling, the one-dimensional convolutional network extracts the time domain waveform features, denoted as X, and uses the extracted time domain waveform features as the input of the gated recurrent network; Each gate control unit receives a time domain waveform feature as input and obtains the output h through calculation; for the t-th gate control unit, it resets the gate r t The current gate input and the previous gate output h t-1 The linear transformation is added and activated by the sigmoid function; For the tth gate control unit, it updates the gate z t By the current gate input x t and the output h of the previous gate t-1 The linear transformation is added and activated by the sigmoid function; The gate control unit generates the candidate hidden state of the current layer by comprehensively updating the information of the gate and the reset gate. By updating the gate to fuse the candidate state with the historical state, the output h of the gate control unit can be obtained t ; For each gating sequence, take the output h of the last gating unit n As a characterization of waveform characteristics.

3. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 2, characterized in that: Each layer of neurons receives signal input from the neurons in the previous layer, and outputs it to the neurons in the next layer after discrete convolution. Its mathematical expression is as follows: Among them, V i k represents the output of the i-th neuron in the k-th layer, W is the one-dimensional convolution kernel, b is the bias, and σ(·) is the convolution activation function.

4. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 3, characterized in that: The pooling method selected is maximum pooling, which retains the local maximum value by sliding the window. The pooling window size is set to j, and the maximum pooling P j as follows: P j =max(V i,s ,V i,s+1 ,...,V i,s+j ) Among them, V i,s ,V i,s+1 ,...,V i,s+j for each element in the pooling window.

5. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 2, characterized in that: The reset gate r of the tth gate control unit t By the current gate input X t and the output h of the previous gate t-1 The linear transformation is added and the sigmoid function is used to activate the reset gate r t : r t =sigmoid(W reset X t +U reset h t-1 +b reset ) Among them, W reset , U reset is the weight matrix of the reset gate, b reset is the offset.

6. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 2, characterized in that: The update gate z of the tth gating unit t By the current gate input x t and the output h of the previous gate t-1 The linear transformation is added and activated with the sigmoid function: z t =sigmoid(W update X t +U update h t-1 +b update ) Among them, W update , U update is the weight matrix of the update gate, b update is the offset.

7. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 1, characterized in that: Candidate hidden states for the t-th gated unit for: Among them, W hidden , U hidden is the weight matrix of the candidate hidden state, b hidden is the bias, and ⊙ is the element-by-element multiplication.

8. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 2, characterized in that: The output h of the tth gating unit t for: Among them, z t is the update gate of the t-th gate control unit, h t-1 is the output of the t-1th gate unit, ⊙ is the element-by-element multiplication, is the candidate hidden state of the t-th gated unit.

9. The method for partial discharge waveform classification based on convolutional gated recurrent network according to claim 1, characterized in that: The cross entropy loss function is used to train the one-dimensional convolutional network and the gated recurrent network. The loss function is as follows: in, is the symbolic function, P ic The network calculates the probability that the i-th sample belongs to the c-th discharge type during training, M is the total number of discharge types, L i is the loss value of the i-th sample, and N is the total number of samples.

10. A partial discharge waveform classification system based on convolutional gated recurrent network, characterized in that: include: The acquisition module receives the electromagnetic waves radiated by partial discharge when a defect occurs in the equipment; The recognition module couples the electromagnetic waves radiated by partial discharge into time domain waveform signals, and then passes them through a one-dimensional convolutional network and a gated recurrent network to realize partial discharge defect recognition.

Images