Fault identification method and system after high-proportion distributed power supply is connected to power distribution network, computer equipment and storage medium

Abstract

The invention discloses a fault identification method and system after a high-proportion distributed power supply is connected to a power distribution network, computer equipment and a storage medium. The method comprises the following steps: acquiring fault data of the high-proportion distributed power supply connected to power distribution networks with different topological structures; extracting fault features; constructing a fault identification model based on a convolutional neural network; integrating a channel attention mechanism into a fault identification model based on a convolutional neural network; training a fault identification model integrated with a channel attention mechanism; testing the trained fault identification model; and after the test is completed, identifying a fault after the high-proportion distributed power supply is connected to the power distribution network. According to the method, the identification model integrated with the channel attention mechanism can be trained through the fault data so as to improve the fault identification accuracy.

Description

Technical Field

[0001] The present application belongs to the technical field of power grid data processing, and specifically relates to a fault identification method, system, computer equipment and storage medium after a high proportion of distributed power sources are connected to a distribution network. Background Art

[0002] At present, the contradiction between the tight energy supply situation and environmental protection is becoming more urgent. Renewable clean energy has replaced non-renewable energy as a new power generation model. The number of distributed power sources (DG) led by wind power and photovoltaics connected to the distribution network has increased rapidly. The distribution network has continuously connected to distributed new energy, various types of energy storage devices and flexible loads such as electric vehicles at a high proportion, prompting the distribution network to transform from a traditional receiving power grid to a new power exchange system that integrates power collection, transmission, storage and distribution.

[0003] Due to the access of a large number of distributed power sources and the introduction of a large number of power electronic equipment, the distribution network will enter a new era of power electronics and become a high-dimensional nonlinear, low-inertia, and low-damping active network. Its operating state will also show strong randomness and disturbance diversity, which will profoundly affect the fault characteristics of the distribution network and greatly increase the difficulty of fault identification. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present application provides a method, system, computer equipment and storage medium for fault identification after a high proportion of distributed power sources are connected to the distribution network. By analyzing the fault characteristics after a high proportion of distributed power sources are connected to the distribution network, time domain, frequency domain and time-frequency domain multi-dimensional fault characteristics are extracted, and these processed data are used to train a fault identification model based on a convolutional neural network. Finally, the trained model is used for fault identification, and artificial intelligence methods are fully utilized to improve the accuracy of fault identification.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions: A fault identification method after a high proportion of distributed power sources are connected to a distribution network, comprising: Obtain fault data of high-proportion distributed generation connected to distribution networks with different topologies; The fault data is divided into a training set and a test set, and the fault features of the training set and the fault features of the test set are extracted; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; Construct a fault identification model based on convolutional neural network; Incorporating a channel attention mechanism into the convolutional neural network-based fault identification model; Using the fault features of the training set to train a fault identification model that incorporates a channel attention mechanism; Using the fault characteristics of the test set to test the trained fault identification model; After the test is completed, the trained fault identification model is used to identify faults after a high proportion of distributed power sources are connected to the distribution network.

[0006] Furthermore, the extraction of the time domain features includes: Using the time domain waveform, three time domain characteristic indicators, peak-to-peak value, standard deviation and kurtosis, are calculated.

[0007] Furthermore, the extraction of the frequency domain features includes: Sampling time domain signals; The sampled time domain signal is used as input and the FFT algorithm is applied to convert the time domain signal into a frequency domain signal, where the frequency domain signal represents the amplitude and phase of the signal at different frequencies; Obtain frequency domain results, perform spectrum analysis, and calculate the center of gravity frequency as the frequency domain characteristic index; the calculation formula is as follows: ; in, is the frequency value of the kth frequency component, is the amplitude of the kth frequency component, and N is the number of frequency components.

[0008] Furthermore, the extraction of the time-frequency domain features includes: Sampling time domain signals; The sampled time domain signal is used as input and decomposed by the VMD algorithm. The VMD algorithm decomposes the time domain signal into multiple modal components through an iterative process. Each modal component represents the vibration mode of the signal at different frequencies and times. The number of functions of the decomposed modes is set to 5. The instantaneous frequency of each IMF component is obtained through Hilbert transform, and the average frequency of each IMF component is taken as the fault time-frequency feature.

[0009] Furthermore, in the fault identification model based on convolutional neural network, the fault identification model based on convolutional neural network includes 6 layers of CNN framework, including 2 convolution layers, 2 pooling layers, 1 fully connected layer and 1 output layer; the loss value optimization algorithm adopts Adam, and the learning rate is set to 1e-4; the loss function adopts the cross entropy loss function; the convolution kernel size adopted is 3×3; the pooling layer adopts average pooling, the size of the pooling window is 2×2, and the stride is 2; the Relu activation function is used for the output of the convolution layer; and the softmax activation function is used for multi-classification tasks in the output layer.

[0010] Furthermore, the channel attention mechanism includes: Perform global average pooling on the convolution output to obtain global information of each channel; The pooled features are input into a fully connected layer to reduce the feature dimension and learn the dependencies between channels; After passing through a fully connected layer, the weight of each channel is output, and the weight is 0-1; The weights are rescaled to the same dimension as the convolution output and multiplied with the original features to adjust the importance of the channels.

[0011] Furthermore, the fault identification model trained by testing the fault characteristics of the test set includes: Step 1: Select the appropriate number and connection method of convolutional layers and pooling layers; Step 2: Initialize the convolutional neural network parameters; Step 3: Input the training data into the convolutional neural network in batches, obtain the diagnostic category of the sample from the output through forward propagation, and calculate the loss; Step 4: Use the back propagation algorithm and Adam optimization function to update the weight parameters of the neural network model; Step 5: Repeat steps 3 and 4 until the accuracy and loss are met or the number of iterations is reached to complete the model training.

[0012] In addition, the present application also provides a fault identification system after a high proportion of distributed power sources are connected to a distribution network, including: An acquisition module is used to obtain fault data of a high proportion of distributed power sources connected to distribution networks with different topological structures; An extraction module, used to divide the fault data into a training set and a test set, and extract the fault features of the training set and the fault features of the test set; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; Model building module, used to build a fault identification model based on convolutional neural network; An attention mechanism integration module, used for integrating the channel attention mechanism into the fault identification model based on the convolutional neural network; A training module, used for training a fault identification model incorporating a channel attention mechanism using the fault features of the training set; A testing module, used to test the trained fault identification model using the fault characteristics of the test set; The identification module is used to use the trained fault identification model to identify faults after a high proportion of distributed power sources are connected to the distribution network.

[0013] In addition, the present application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-described method when executing the computer program.

[0014] In addition, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the method described above is implemented.

[0015] Compared with the prior art, this application has the following advantages: The present application includes three parts: fault feature analysis and extraction after a high proportion of distributed power sources are connected to the distribution network, and fault identification model training based on convolutional neural networks: first, different fault data under different conditions of the actual distribution network are collected, or data of different faults occurring in different conditions of the line model built in software such as PSCAD / Simulink are recorded, and the impact of distributed power access on the fault characteristics is analyzed from two aspects: the waveform and harmonic content of the fault characteristics. Then, based on the change of fault characteristics, 9 fault feature indicators in the time domain, frequency domain and time-frequency domain are extracted, and different labels are marked for faults occurring at different locations; finally, the processed data is divided into a training set and a test set, the basic structure of the convolutional neural network model is set, and the fault identification model is trained until it is completed or the required accuracy is reached.

[0016] Due to the impact of distributed power generation on multiple aspects of fault characteristics after access, including steady-state characteristics, transient characteristics, harmonic distribution, etc., this application analyzes the fault characteristics after access to distributed power sources for different distribution network topologies: radial type, three-segment two-connection type, and ring network, and extracts multi-dimensional fault features in time domain, frequency domain, and time-frequency domain, making feature extraction more comprehensive, and adding weights to important features by adding an attention mechanism to improve recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application but do not constitute an improper limitation on the present application. In the drawings: Figure 1 A flow chart of the method of this application; Figure 2 The following is a comparison of fault characteristics before and after the distributed generation is connected, where (a) is the amplitude and shape of the three-phase current waveform at the fault point before the distributed generation is connected, (b) is the amplitude and shape of the three-phase current waveform at the fault point after the distributed generation is connected, (c) is the harmonic content of the current at the fault point before the distributed generation is connected, and (d) is the harmonic content of the current at the fault point after the distributed generation is connected; Figure 3This is a schematic diagram of the radial distribution network structure; Figure 4 This is a schematic diagram of the structure of a three-segment two-connection distribution network; Figure 5 This is a schematic diagram of the ring distribution network structure; Figure 6 This is the system structure diagram of this application; Figure 7 A schematic diagram of the structure of the computer device of the present application; Figure 8 This is a schematic diagram of the structure of another computer device of the present application. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0019] In the description of this application, it should be understood that the following description of this application is only understood as a description of individual embodiments of the technical solution of this application. Other embodiments are not reflected in the following description, but it does not mean that this application excludes these other embodiments, and the technical solution of this application is not limited to the specific implementation methods described below, and the protection scope of this application is not limited to only the specific implementation methods described below. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of this application.

[0020] It should be noted that if the terms "first", "second", etc. appear in the specification and claims of the present application and the above-mentioned drawings, the description is only used to distinguish similar objects, and is 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 of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising 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.

[0021] In some embodiments, Figure 1 As shown, a fault identification method after a high proportion of distributed power sources are connected to a distribution network includes: S1: Obtain fault data of high-proportion distributed generation connected to distribution networks with different topologies; Specifically, when distributed power sources are connected to the grid, the grid structure changes from radial to multi-power type. Distributed power sources with small capacity and small number have limited impact on fault current, while distributed power sources with large capacity and concentrated distribution have a significant impact on the magnitude of fault current and may even change the direction of fault current.

[0022] At the same time, distributed photovoltaics are connected to the grid through DC-to-AC inverters, and their power electronic converters generate a large amount of harmonics, which seriously affect the fault characteristics.

[0023] It is possible to collect different fault data under different conditions in the actual distribution network, or record the data of different faults occurring in different situations in the line model built in software such as PSCAD / Simulink, and perform Fourier decomposition on the data to obtain the harmonic content.

[0024] Compare and analyze different fault waveforms and harmonics of the distributed power supply without access and the distributed power supply at different locations, such as Figure 2 Comparison of fault characteristics before and after distributed power supply is connected.

[0025] The following is an analysis of the fault characteristics of three distribution networks with different topologies: The characteristics of each ground fault and short circuit fault of the radial distribution network are analyzed from two aspects: steady-state characteristics and transient characteristics. In these two cases, different neutral point grounding methods and different fault locations are analyzed from multiple angles. Then, the characteristics of two typical distribution network structures, the three-segment two-connection distribution network and the ring distribution network, are analyzed in detail when a fault occurs. The steady-state and transient fault characteristics of the system are calculated under various neutral point grounding methods and different fault line conditions. At the same time, the influence of distributed power access on fault current and harmonic distribution can be analyzed, and the analysis results are verified by Matlab / Simulink simulation model, providing a reference for the subsequent extraction of characteristic indicators.

[0026] In radial distribution networks (such as Figure 3 As shown in the figure, the access of a high proportion of distributed generation (DG) significantly affects the fault characteristics. The access of DG increases the fault current amplitude, changes the flow direction of the fault current, and complicates the traditional unidirectional current flow characteristics. The power electronic equipment such as the inverter of DG introduces a large amount of harmonic current, resulting in a significant increase in the harmonic content in the distribution network, which seriously affects the power quality and system stability. The access of distributed generation makes the flow direction complicated. It is no longer a unidirectional flow from the power source to the load, but may flow in two directions or in multiple directions, which increases the uncertainty of the fault current path. Due to the existence of distributed generation, the fault current may propagate through multiple paths, making fault identification more difficult.

[0027] In the three-segment two-connection distribution network (such as Figure 4 In the figure, the access of a high proportion of distributed generation (DG) has the following main effects on the fault characteristics: First, the access of distributed generation will increase the current transmission between the segments, which may cause some segments to carry higher current loads, resulting in uneven current distribution. Second, the access of distributed generation will change the voltage distribution within and between segments. Distributed generation usually converts DC power into AC power through inverters and connects to the distribution network. Its voltage level and frequency may not be completely consistent with traditional generators or transmission networks, which leads to increased voltage fluctuations near the access point. Third, the access of distributed generation significantly increases the load on the interconnection lines between segments.

[0028] In a ring distribution network (such as Figure 5 In the example (shown in Figure 1), the access of a high proportion of distributed generation (DG) has the following main effects on the fault characteristics: First, the access of distributed generation may affect the voltage and frequency stability of the system, and this effect may be more significant in the ring distribution network. Second, the characteristics of the ring distribution network determine that under fault conditions, the current can propagate along the loop multiple times. The access of distributed generation increases the total current input of the ring distribution network, especially in the case of a fault, which may lead to a higher fault current.

[0029] The above analysis can better understand the fault characteristics of distribution networks with various topologies after connecting to a high proportion of distributed power sources.

[0030] Three fault occurrence models with different topologies were built through PSCAD. 3000 waveform data were collected for each topology model, including 3 different faults at various locations. These three faults were single-phase short circuit fault, two-phase short circuit fault and three-phase short circuit fault. They were short circuit faults occurring on different feeders of DG and before and after DG on the same feeder. Each fault condition contained 1000 data under each topology, and each fault condition had 3000 data, for a total of 9000 data. The labels of these three fault conditions were recorded as 0-2 respectively.

[0031] S2: dividing the fault data into a training set and a test set, and extracting the fault features of the training set and the fault features of the test set; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; This application is applicable to different distribution network topologies, reflecting the generalization of the identification model. The high proportion of distributed power access leads to complex fault signal characteristics. In order to overcome this difficulty, this application proposes to extract multi-dimensional fault features in the time domain, frequency domain, and time-frequency domain, making feature extraction more comprehensive.

[0032] Specifically, the extraction of the time domain features includes: Using the time domain waveform, three time domain characteristic indicators, peak-to-peak value, standard deviation and kurtosis, are calculated.

[0033] The peak-to-peak value refers to the difference between the maximum and minimum values ​​of a signal within a complete cycle. This indicator is often used to describe the dynamic range of a signal and its amplitude. For non-periodic signals, the peak-to-peak value refers to the difference between the maximum and minimum values ​​within the entire signal range.

[0034] Standard deviation is a statistical indicator used to measure the dispersion or volatility of values ​​in a set of data. When the standard deviation is larger, it means that the volatility and dispersion of the data are higher; the smaller the standard deviation, the lower the volatility of the data. The calculation formula of standard deviation is as follows: ; Where N is the number of observations in the dataset, It is Observations, is the average value.

[0035] Kurtosis is used to measure the sharpness or flatness of data distribution. The calculation formula of kurtosis can be implemented using various statistical software or programming languages.

[0036] Specifically, the frequency domain feature extraction can use the fast Fourier transform (FFT) to convert the fault features in the time domain to the frequency domain for analysis, and convert the time domain signal into a frequency domain signal. The following are the basic steps for converting the time domain signal into the frequency domain signal using the fast Fourier transform (FFT), including: Sampling time domain signals; Determine the sampling rate, which is the number of samples collected per second. The sampling rate determines the frequency resolution of the signal; Choose an appropriate FFT length. The FFT length is usually chosen to be a power of 2. The sampled time domain signal is used as input and the FFT algorithm is applied to convert the time domain signal into a frequency domain signal, where the frequency domain signal represents the amplitude and phase of the signal at different frequencies; Obtain frequency domain results, perform spectrum analysis, and calculate the centroid frequency as a frequency domain characteristic indicator; the centroid frequency is the center position of the spectrum distribution. It is the weighted average frequency of each frequency component in the spectrum, reflecting the contribution and distribution of each frequency component in the spectrum; the calculation formula is as follows: ; in, is the frequency value of the kth frequency component, is the amplitude of the kth frequency component, and N is the number of frequency components.

[0037] Specifically, the extraction of the time-frequency domain features can use the variational mode decomposition (VMD) method to convert the fault features in the time domain to the time-frequency domain. VMD is a signal processing method used to convert time domain signals into time-frequency domain representations. Unlike the traditional Fourier transform method, VMD is an adaptive decomposition technology that can adapt to the characteristics of nonlinear and non-stationary signals. The following are the basic steps for converting time domain signals into time-frequency domain representations using VMD, including: Sampling time domain signals; Determine the parameters of VMD decomposition, including signal sampling rate, number of decomposition layers, regularization parameters, etc.; The sampled time domain signal is used as input and decomposed by the VMD algorithm. The VMD algorithm decomposes the time domain signal into multiple modal components through an iterative process. Each modal component represents the vibration mode of the signal at different frequencies and times. The number of functions of the decomposed modes is set to 5. The instantaneous frequency of each IMF component is obtained through Hilbert transform, and the average frequency of each IMF component is taken as the fault time-frequency feature.

[0038] Table 1 below is a description of multi-dimensional fault feature quantities: Table 1 Description of multi-dimensional fault characteristics

[0039] In some embodiments, the 9000 data collected above can be divided into a training set and a test set at a ratio of 9: 1. Feature processing is performed on the collected data, and the time domain waveform is converted to the frequency domain and the time-frequency domain through the above transformation method, and then the 9-dimensional fault feature is calculated.

[0040] S3: Construct a fault identification model based on convolutional neural network; Specifically, convolutional neural network (CNN) is a deep learning model that has been widely used in many fields such as image recognition, computer vision, and natural language processing. CNN mainly consists of the following parts: convolution layer, pooling layer, activation function, and fully connected layer.

[0041] In the fault identification model based on convolutional neural network, the fault identification model based on convolutional neural network includes 6 layers of CNN framework, including 2 convolution layers, 2 pooling layers, 1 fully connected layer and 1 output layer; Adam is used as the loss value optimization algorithm, and the learning rate is set to 1e-4; the loss function adopts the cross entropy loss function; the convolution kernel size adopted is 3×3; the pooling layer adopts average pooling, the size of the pooling window is 2×2, and the stride is 2; the Relu activation function is used for the output of the convolution layer; and the softmax activation function is used for multi-classification tasks in the output layer.

[0042] S4: Integrating the channel attention mechanism into the fault identification model based on the convolutional neural network; The Channel Attention Mechanism is a technique for enhancing the feature expression capability in deep neural networks, and is mainly used in convolutional neural networks (CNNs). It assigns different weights to different channels of the input feature map, allowing the network to focus more on those feature channels that are more critical in specific tasks, thereby effectively improving the model's expression capability and overall performance.

[0043] The channel attention mechanism includes: Perform global average pooling on the convolution output to obtain global information of each channel; The pooled features are input into a fully connected layer to reduce the feature dimension and learn the dependencies between channels; After passing through a fully connected layer, the weight of each channel is output, and the weight is 0-1; The weights are rescaled to the same dimension as the convolution output and multiplied with the original features to adjust the importance of the channels.

[0044] Specifically, the corresponding modules of the channel attention mechanism can include: Global Average Pooling: Perform global average pooling on the convolution output to obtain global information for each channel. The output shape is (batch_size, channels).

[0045] Fully connected layer I: The pooled features are input into a fully connected layer (usually using ReLU activation) to reduce the feature dimension and learn the dependencies between channels.

[0046] Fully connected layer II: After another fully connected layer (usually using Sigmoid activation), the weight of each channel is output (between 0 and 1). The output shape is (batch_size, channels).

[0047] Scaling channels: These weights are rescaled to the same dimension as the convolution output and multiplied with the original features, adjusting the importance of the channels.

[0048] By introducing the channel attention mechanism after the second convolutional layer of the convolutional neural network built above, the network can automatically identify and enhance the feature channels that are more important for specific tasks, while suppressing unimportant channels. This design can improve the performance of the model, especially in tasks such as image processing, object detection, and classification.

[0049] S5: using the fault features of the training set to train a fault identification model that incorporates a channel attention mechanism; Specifically, the fault identification model trained by testing the fault characteristics of the test set includes: Step 1: Select the appropriate number and connection method of convolutional layers and pooling layers; Step 2: Initialize the convolutional neural network parameters; Step 3: Input the training data into the convolutional neural network in batches, obtain the diagnostic category of the sample from the output through forward propagation, and calculate the loss; Step 4: Use the back propagation algorithm and Adam optimization function to update the weight parameters of the neural network model; Step 5: Repeat steps 3 and 4 until the accuracy and loss are met or the number of iterations is reached to complete the model training.

[0050] S6: using the fault characteristics of the test set to test the trained fault identification model; S7: After the test is completed, the trained fault identification model is used to identify faults after a high proportion of distributed power sources are connected to the distribution network.

[0051] This application is applicable to scenarios with high complexity and difficulty in identifying distribution network scenarios with high proportion of distributed power access. The methods for extracting features in the three dimensions are also different, especially the features in the time-frequency domain use the VMD algorithm to first decompose the signal, and then use the Hilbert transform to find the instantaneous frequency of each component as a feature. In addition, to avoid the method of judging by indicators in a single time-frequency domain, a multi-index feature self-learning and intelligent discrimination method of CNN is established, and a key sliding window attention mechanism is added to the model to avoid overfitting and time-consuming training of CNN, and improve the recognition accuracy under small samples.

[0052] In some embodiments, Figure 6 The present application also provides a fault identification system 1 after a high proportion of distributed power sources are connected to a distribution network, comprising: An acquisition module 11 is used to acquire fault data of a high proportion of distributed power sources connected to distribution networks with different topological structures; An extraction module 12 is used to divide the fault data into a training set and a test set, and extract the fault features of the training set and the fault features of the test set; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; A model building module 13, used to build a fault identification model based on a convolutional neural network; An attention mechanism integration module 14 is used to integrate the channel attention mechanism into the fault identification model based on the convolutional neural network; A training module 15 is used to train a fault identification model incorporating a channel attention mechanism using the fault features of the training set; A testing module 16, used to test the trained fault identification model using the fault characteristics of the test set; The identification module 17 is used to use the trained fault identification model to identify faults after a high proportion of distributed power sources are connected to the distribution network.

[0053] Each module in the above-mentioned fault identification system after a high proportion of distributed power sources is connected to the distribution network can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0054] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a fault identification method after a high proportion of distributed power sources are connected to a distribution network is implemented.

[0055] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 8As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a fault identification method after a high proportion of distributed power sources are connected to a distribution network is implemented. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse, etc.

[0056] Those skilled in the art will understand that Figure 7 and Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0057] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0058] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0059] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0060] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0061] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in any type of volatile or non-volatile storage device, or a combination thereof. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache.By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDR AM), enhanced synchronous dynamic random access memory (ESDRAM Enhanced Synchronous Dynamic Random Access Memory), synchronous link dynamic random access memory (SLDRAM), direct memory bus random access memory (DRRAM). The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0062] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A fault identification method after a high proportion of distributed power sources are connected to a distribution network, characterized in that: include: Obtain fault data of high-proportion distributed generation connected to distribution networks with different topologies; The fault data is divided into a training set and a test set, and the fault features of the training set and the fault features of the test set are extracted; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; Construct a fault identification model based on convolutional neural network; Incorporating a channel attention mechanism into the convolutional neural network-based fault identification model; Using the fault features of the training set to train a fault identification model that incorporates a channel attention mechanism; Using the fault characteristics of the test set to test the trained fault identification model; After the test is completed, the trained fault identification model is used to identify the faults after a high proportion of distributed power sources are connected to the distribution network; The extraction of the time domain features includes: Using the time domain waveform, calculate the three time domain characteristic indicators of peak-to-peak value, standard deviation and kurtosis; The extraction of the frequency domain features includes: Sampling time domain signals; The sampled time domain signal is used as input and the FFT algorithm is applied to convert the time domain signal into a frequency domain signal, where the frequency domain signal represents the amplitude and phase of the signal at different frequencies; Obtain frequency domain results, perform spectrum analysis, and calculate the center of gravity frequency as the frequency domain characteristic index; the calculation formula is as follows: ； in, is the frequency value of the kth frequency component, is the amplitude of the kth frequency component, N is the number of frequency components, is the acceleration due to gravity; The extraction of the time-frequency domain features includes: Sampling time domain signals; The sampled time domain signal is used as input and decomposed by the VMD algorithm. The VMD algorithm decomposes the time domain signal into multiple modal components through an iterative process. Each modal component represents the vibration mode of the signal at different frequencies and times. The number of functions of the decomposed modes is set to 5. The instantaneous frequency of each IMF component is obtained through Hilbert transform, and the average frequency of each IMF component is taken as the fault time-frequency feature.

2. The fault identification method after a high proportion of distributed power sources is connected to a distribution network according to claim 1, characterized in that: In the construction of the fault identification model based on convolutional neural network, the fault identification model based on convolutional neural network includes 2 convolutional layers, 2 pooling layers, 1 fully connected layer and 1 output layer, a total of 6 layers of CNN framework; Adam is used as the loss value optimization algorithm, and the learning rate is set to 1e-4; The loss function uses the cross entropy loss function; the convolution kernel size used is 3×3; the pooling layer uses average pooling, the size of the pooling window is 2×2, and the stride is 2; the Relu activation function is used for the output of the convolution layer; the softmax activation function is used for multi-classification tasks in the output layer.

3. The fault identification method after a high proportion of distributed power sources is connected to a distribution network according to claim 1, characterized in that: The channel attention mechanism includes: Perform global average pooling on the convolution output to obtain global information of each channel; The pooled features are input into a fully connected layer to reduce the feature dimension and learn the dependencies between channels; After passing through a fully connected layer, the weight of each channel is output, and the weight is 0-1; The weights are rescaled to the same dimension as the convolution output and multiplied with the original features to adjust the importance of the channels.

4. The fault identification method after a high proportion of distributed power sources is connected to a distribution network according to claim 1, characterized in that: The fault identification model trained by testing the fault characteristics of the test set includes: Step 1: Select the appropriate number and connection method of convolutional layers and pooling layers; Step 2: Initialize the convolutional neural network parameters; Step 3: Input the training data into the convolutional neural network in batches, obtain the diagnostic category of the sample from the output through forward propagation, and calculate the loss; Step 4: Use the back propagation algorithm and Adam optimization function to update the weight parameters of the neural network model; Step 5: Repeat steps 3 and 4 until the accuracy and loss are met or the number of iterations is reached to complete the model training.

5. A fault identification system after a high proportion of distributed power sources are connected to a distribution network, characterized in that: include: An acquisition module is used to obtain fault data of a high proportion of distributed power sources connected to distribution networks with different topological structures; An extraction module, used to divide the fault data into a training set and a test set, and extract the fault features of the training set and the fault features of the test set; the fault features of the training set and the fault features of the test set include time domain features, frequency domain features and time-frequency domain features; Model building module, used to build a fault identification model based on convolutional neural network; A mechanism integration module, used for integrating the channel attention mechanism into the fault identification model based on the convolutional neural network; A training module, used for training a fault identification model incorporating a channel attention mechanism using the fault features of the training set; A testing module, used to test the trained fault identification model using the fault characteristics of the test set; The identification module is used to use the trained fault identification model to identify faults after a high proportion of distributed power sources are connected to the distribution network.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images