Power distribution network fault type identification method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to the technical field of power distribution network fault research and judgment, and provides a power distribution network fault type identification method and device, computer equipment, a readable storage medium and a program product. The method comprises the following steps: performing feature extraction on real-time situation awareness data, equipment operation risk data and meteorological data of a fault position of the power distribution network to obtain fault feature data corresponding to the fault position of the power distribution network; performing multi-scale wavelet decomposition on the fault feature data to obtain a fault feature vector under each scale; and inputting the fault feature vector into a pre-constructed power distribution network fault type identification model to obtain a fault type corresponding to the fault position of the power distribution network. By adopting the method, the fault features under various scales can be extracted and integrated, the change features of the power distribution network fault on different time scales and space scales are effectively captured, and the accuracy and robustness of power distribution network fault type identification are improved.

Description

Technical Field

[0001] This application relates to the field of power distribution network fault assessment technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for identifying power distribution network fault types. Background Technology

[0002] Accurate and rapid diagnosis of short-circuit faults in distribution networks is of paramount importance for improving power supply reliability and ensuring the safe operation of the distribution network.

[0003] Currently, fault type identification in distribution networks is mainly based on information such as current and voltage, or on the operating status of protective switches and circuit breakers. However, traditional methods for identifying distribution network fault types rely on relatively limited information sources, resulting in low accuracy when faced with complex situations. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, device, computer equipment, computer-readable storage medium, and computer program product for identifying distribution network fault types to address the aforementioned technical problems.

[0005] Firstly, this application provides a method for identifying distribution network fault types, including:

[0006] Feature extraction is performed on real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network.

[0007] The fault feature data is subjected to multi-scale wavelet decomposition to obtain the fault feature vector at each scale.

[0008] The fault feature vector is input into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

[0009] In one embodiment, before extracting features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network, the method further includes:

[0010] A fault location model for the distribution network is trained based on the backpropagation neural network model.

[0011] The operation information of each protection device and each circuit breaker in the distribution network is input into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network, so as to determine the fault location of the distribution network.

[0012] In one embodiment, training the distribution network fault location model based on the backpropagation neural network model includes:

[0013] Initialize the weight coefficients and thresholds of the hidden and output layers of the backpropagation neural network;

[0014] Training samples are obtained based on the action information samples of each protection device and each circuit breaker in the distribution network.

[0015] At least one round of training is performed, wherein in each round of training, the training samples are input into the backpropagation neural network to obtain the input data and output data of the hidden layer and output layer of the backpropagation neural network for the current round; based on the output data of the output layer of the backpropagation neural network for the current round, the output error of the training samples for the current round is obtained; based on the output error of the training samples for the current round, the weight coefficients and thresholds of the hidden layer and output layer for the next round are updated.

[0016] If the current output error of the training sample is greater than a preset threshold, then the next round of training is performed; if the current output error of the training sample is less than or equal to the preset threshold, then training is stopped, and the distribution network fault location model is obtained based on the weight coefficients and thresholds of the hidden layer and the output layer for the next round.

[0017] In one embodiment, the feature extraction of real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network includes:

[0018] Real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network are preprocessed to obtain preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data.

[0019] The mean and standard deviation of preprocessed real-time situational awareness data, equipment operation risk data and meteorological data are extracted to obtain the time-frequency characteristics corresponding to the fault location of the distribution network.

[0020] Extract the spectral density values ​​of preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the frequency domain characteristics corresponding to the fault location of the distribution network;

[0021] Based on the time-frequency and frequency-domain characteristics corresponding to the fault location in the distribution network, fault characteristic data corresponding to the fault location in the distribution network are obtained.

[0022] In one embodiment, the step of performing multi-scale wavelet decomposition on the fault feature data to obtain fault feature vectors at each scale includes:

[0023] The fault feature data is subjected to multi-scale wavelet decomposition to obtain fault feature sub-data at different scales;

[0024] Based on the fault feature sub-data at each scale, the fault feature vector at each scale is obtained.

[0025] In one embodiment, the step of inputting the fault feature vector into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network includes:

[0026] The fault feature vector at each scale is input into different feature matchers in the pre-built distribution network fault type identification model to match the fault feature vector with the fault type, thereby obtaining the corresponding matching result set for each scale.

[0027] By statistically analyzing the matching result set for each scale, the voting statistics for each fault type are obtained.

[0028] If the voting statistics corresponding to one of the fault types exceed the voting statistics threshold, then the fault type is determined as the fault type corresponding to the fault location of the distribution network.

[0029] Secondly, this application also provides a device for identifying the type of fault in a distribution network, comprising:

[0030] The fault feature data acquisition module is used to extract features from real-time situational awareness data, equipment operation risk data and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network.

[0031] Fault feature vector acquisition is used to perform multi-scale wavelet decomposition on the fault feature data to obtain the fault feature vector at each scale.

[0032] The fault type identification module is used to input the fault feature vector into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

[0033] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the above-described method.

[0034] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, the computer program being executed by a processor using the methods described above.

[0035] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that is executed by a processor using the methods described above.

[0036] The aforementioned methods, devices, computer equipment, computer-readable storage media, and computer program products for identifying distribution network fault types extract features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location. Multi-scale wavelet decomposition is performed on the fault feature data to obtain fault feature vectors at each scale. These fault feature vectors are then input into a pre-constructed distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network. This application extracts features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location, thereby obtaining fault feature vectors at each scale and ultimately determining the fault type. This approach can extract and integrate fault features at multiple scales, effectively capturing the changing characteristics of distribution network faults at different time and spatial scales, and improving the accuracy and robustness of distribution network fault type identification. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is an application environment diagram of a method for identifying distribution network fault types in one embodiment;

[0039] Figure 2 This is a flowchart illustrating a method for identifying distribution network fault types in one embodiment;

[0040] Figure 3 This is a schematic diagram of the learning process of a backpropagation neural network model in one embodiment;

[0041] Figure 4 This is a structural block diagram of a power distribution network fault type identification device in one embodiment;

[0042] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0044] It should be noted that the terms "comprising" and "having," and any variations thereof, as used in this application, are intended to cover non-exclusive inclusion. The term "multiple" as used in this application refers to two or more. The term "and / or" as used in this application refers to one of the solutions, or any combination of multiple solutions.

[0045] The method for identifying distribution network fault types provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on the cloud or other network servers. Terminal 102 extracts features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location, thus obtaining a fault feature vector at each scale. The fault feature vector is then input into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, projection devices, etc. Server 104 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0046] In one exemplary embodiment, such as Figure 2 As shown, a method for identifying fault types in a distribution network is provided, which can be applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps S201 to S203. Wherein:

[0047] Step S201: Extract features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network.

[0048] Real-time situational awareness data refers to real-time information collected through various monitoring devices that reflects the operating status and characteristics of the power distribution network.

[0049] Equipment operation risk data refers to data that reflects the health status of equipment, the probability of failure, and environmental influencing factors.

[0050] Meteorological data refers to environmental parameters of the power distribution network and the location of equipment within it.

[0051] A distribution network fault location model can be constructed based on the backpropagation neural network model to achieve accurate fault location in the distribution network. After accurately locating the fault location, features are extracted from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location to obtain the time-frequency features and frequency domain features corresponding to the fault location in the distribution network, which serve as the fault feature data corresponding to the fault location in the distribution network.

[0052] Step S202: Perform multi-scale wavelet decomposition on the fault feature data to obtain the fault feature vector at each scale.

[0053] Each feature in the fault feature data can be decomposed into multiple scales to obtain sub-features at multiple scales corresponding to each feature. Sub-features at the same scale can be combined to obtain the fault feature vector at each scale.

[0054] Step S203: Input the fault feature vector into the pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

[0055] A distribution network fault type identification model can be trained based on fault feature vector sample data and normal feature vector sample data, which can then be used as a pre-built distribution network fault type identification model.

[0056] The fault feature vector can be input into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

[0057] In the above-mentioned method for identifying distribution network fault types, features are extracted from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network. This yields fault feature vectors at each scale, thereby obtaining the fault type corresponding to the fault location in the distribution network. This method can extract and integrate fault features at multiple scales, effectively capturing the changing characteristics of distribution network faults at different time and spatial scales, and improving the accuracy and robustness of distribution network fault type identification.

[0058] In one embodiment, before extracting features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network, the method provided in this application further includes: training a distribution network fault location model based on a backpropagation neural network model; inputting the action information of each protection device and the action information of each circuit breaker in the distribution network into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network, so as to determine the fault location in the distribution network.

[0059] A fault location model for the power distribution network can be trained based on the backpropagation neural network model.

[0060] The operation information of each protection device and each circuit breaker in the distribution network can be input into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network; the fault location of the distribution network can be determined based on the fault probability of each protection device and each circuit breaker in the distribution network.

[0061] In this embodiment, a distribution network fault location model is trained based on the backpropagation neural network model. Utilizing the powerful adaptive and generalization capabilities of the backpropagation neural network, a distribution network fault location model with good performance is obtained. The action information of each protection device and each circuit breaker in the distribution network is input into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network, thereby determining the fault location of the distribution network. This can solve the problems of false operation, failure to operate, and data distortion in the switching information of the distribution network, thus achieving accurate fault location.

[0062] In one embodiment, a distribution network fault location model is trained based on a backpropagation neural network model. The specific steps are as follows: Initialize the weight coefficients and thresholds of the hidden layer and output layer of the backpropagation neural network; obtain training samples based on the action information samples of each protection device and each circuit breaker in the distribution network; perform at least one round of training, wherein in each round of training, the training samples are input into the backpropagation neural network to obtain the input data and output data of the hidden layer and output layer of the backpropagation neural network for the current round; obtain the output error of the training samples for the current round based on the output data of the output layer of the backpropagation neural network for the current round; update the weight coefficients and thresholds of the hidden layer and output layer for the next round based on the output error of the training samples for the current round; if the output error of the training samples for the current round is greater than a preset threshold, then proceed to the next round of training; if the output error of the training samples for the current round is less than or equal to the preset threshold, then stop training, and obtain the distribution network fault location model based on the weight coefficients and thresholds of the hidden layer and output layer for the next round.

[0063] The learning process of a backpropagation neural network model mainly includes two steps: the first step is the forward propagation stage, where information is passed from the hidden layer to the output layer; the second step is to use error negative feedback to adjust the various parameters in the network.

[0064] The weight coefficients (also called weights) and thresholds of the hidden and output layers of the backpropagation neural network can be initialized; training samples can be obtained based on the action information samples of each protection device and each circuit breaker in the distribution network; and at least one round of training can be performed based on the training samples.

[0065] During the i-th round of training, the training samples are... The formulas for calculating the input and output values ​​of the m-th neuron in the hidden layer of the backpropagation neural network are shown in Equations (1) and (2).

[0066] (1)

[0067] (2)

[0068] In the formula, This represents the input value of the hidden layer. This represents the output value of the hidden layer. This represents the weight coefficients between the input layer and the hidden layer. This represents the threshold of the hidden layer. represents the transfer function of the hidden layer, and N represents the number of neurons in the input layer.

[0069] The output data of the hidden layer during the i-th training round can be obtained from the output values ​​of all neurons in the hidden layer during the i-th training round.

[0070] The formulas for calculating the input value and output value received by the a-th neuron in the output layer are shown in equations (3) and (4).

[0071] (3)

[0072] (4)

[0073] In the formula, This represents the input value of the output layer. Indicates the output value of the output layer; This represents the weight coefficients between the output layer and the hidden layer. Indicates the threshold of the output layer. This represents the activation function of the output layer, and S represents the number of neurons in the hidden layer.

[0074] The output data of the output layer in the i-th training round can be obtained from the output values ​​of all neurons in the output layer in the i-th training round.

[0075] The output error of the training samples in the i-th round can be obtained from the output data and error function of the output layer of the backpropagation neural network during the i-th round of training.

[0076] The weight coefficients between the output layer and the hidden layer can be adjusted according to the output error of the i-th round of the training samples, as shown in Equations (5) and (6), so as to obtain the weight coefficients between the output layer and the hidden layer in the (i+1)-th round.

[0077] (5)

[0078] (6)

[0079] In the formula, This represents the weight adjustment amount between the output layer and the hidden layer. These represent the weight adjustments between the hidden layer and the input layer, respectively. represents the model's learning rate, i represents the number of training iterations, and M represents the number of neurons in the output layer sample.

[0080] The calculation formulas for optimizing the thresholds of the hidden layer and the output layer are shown in Equations (7) and (8), thus obtaining the thresholds of the hidden layer and the output layer in the (i+1)th round.

[0081] (7)

[0082] (8)

[0083] In the formula, This represents the threshold of the hidden layer. This represents the threshold of the output layer.

[0084] If the output error of the training sample in the i-th round is greater than a preset threshold, then the (i+1)-th round of training is performed; if the output error of the training sample in the i-th round is less than or equal to the preset threshold, then training is stopped, and the distribution network fault location model is obtained based on the weight coefficients of the hidden layer and the output layer in the (i+1)-th round and the threshold in the (i+1)-th round. The preset threshold can be determined according to the actual situation.

[0085] In this embodiment, a distribution network fault location model is trained based on the backpropagation neural network model. The powerful adaptive and generalization capabilities of the backpropagation neural network can be utilized to obtain a distribution network fault location model with better performance.

[0086] In one embodiment, feature extraction is performed on real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network. The specific steps are as follows: preprocessing the real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data; extracting the mean and standard deviation of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the time-frequency features corresponding to the fault location in the distribution network; extracting the spectral density values ​​of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the frequency domain features corresponding to the fault location in the distribution network; and obtaining the fault feature data corresponding to the fault location in the distribution network based on the time-frequency features and frequency domain features corresponding to the fault location in the distribution network.

[0087] For example, assume that the real-time situational awareness data of the fault location in the distribution network is a one-dimensional signal. , where t represents the time variable. For Preprocessing is performed to obtain the preprocessed signal. As shown in equations (9) and (10), the preprocessed signal is extracted. The mean and standard deviation (std) are used as the signal. The corresponding time-frequency characteristics are shown in equation (11), based on the spectral density function. Extract the preprocessed signal The spectral density value, as the signal The corresponding frequency domain characteristics.

[0088] (9)

[0089] (10)

[0090] (11)

[0091] In the formula, This represents the operations performed on the Fourier transform.

[0092] Based on the signal By analyzing the corresponding time-frequency and frequency-domain characteristics, fault characteristic data corresponding to the fault location in the distribution network can be obtained.

[0093] In this embodiment, the mean and standard deviation of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data are extracted to obtain the time-frequency features corresponding to the fault location of the distribution network; the spectral density values ​​of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data are extracted to obtain the frequency domain features corresponding to the fault location of the distribution network, so as to obtain the fault feature data corresponding to the fault location of the distribution network. Fault features at multiple scales of the fault location of the distribution network can be extracted.

[0094] In one embodiment, the fault feature data is decomposed into a multi-scale wavelet decomposition to obtain a fault feature vector at each scale. The specific steps are as follows: the fault feature data is decomposed into a multi-scale wavelet decomposition to obtain fault feature sub-data at different scales; based on the fault feature sub-data at each scale, the fault feature vector at each scale is obtained.

[0095] For example, assume that the real-time situational awareness data of the fault location in the distribution network is a one-dimensional signal. , for signal The extracted fault feature data is subjected to multi-scale wavelet decomposition to obtain a series of fault feature sub-data at different scales. It can also be called a sub-signal. , where r is used to represent the scale.

[0096] Based on the sub-signals at each scale Calculate the corresponding fault feature vector As shown in equation (12).

[0097] (12)

[0098] In the formula, Represents sub-signal The mean, Represents sub-signal standard deviation Sub-signal The power spectral density function.

[0099] In this embodiment, the fault feature data is decomposed into multi-scale wavelet decomposition to obtain fault feature sub-data at different scales; based on the fault feature sub-data at each scale, the fault feature vector at each scale is obtained, thereby obtaining the fault feature vector at multiple scales of the fault location in the distribution network.

[0100] In one embodiment, the fault feature vector is input into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network. The specific steps are as follows: the fault feature vector at each scale is input into different feature matchers in the pre-built distribution network fault type identification model to match the fault feature vector with the fault type, and the corresponding matching result set at each scale is obtained; the corresponding matching result set at each scale is statistically analyzed to obtain the voting statistics corresponding to each fault type; if the voting statistics corresponding to a fault type exceeds the voting statistics threshold, the fault type is determined as the fault type corresponding to the fault location in the distribution network.

[0101] To improve the accuracy of distribution network fault type identification, the distribution network fault type identification model can be solved based on ensemble learning algorithms.

[0102] Assume that the distribution network fault type identification model has y feature matchers, labeled M1, M2, ..., My. The fault feature vector at each scale can be input into each feature matcher in the pre-built distribution network fault type identification model. Each feature matcher performs a feature matching operation on the fault feature vector at each scale to match the fault feature vector with the fault type, obtaining the matching result corresponding to the fault feature vector at each scale, thus obtaining the corresponding matching result set for each scale.

[0103] To achieve integrated decision-making, a voting mechanism is introduced. This is applied to the fault location of each distribution network. It can be the position coordinates of the sliding window. The system analyzes the outputs of all feature matchers and makes a final decision based on a voting mechanism. More specifically, it analyzes the matching result set for each scale to obtain the voting statistics for each fault type.

[0104] The voting statistics for fault type k at the s-th level scale As shown in equation (13).

[0105] (13)

[0106] In the formula, This indicates that for the s-th layer scale feature matcher Mb at position The location is determined to be a fault type. Similarity score; This indicates an indicator function. The function's value is 1 when the condition within the parentheses is true, and 0 otherwise.

[0107] Within each scale (also known as scale space), it is possible to utilize Obtain the voting statistics matrix for fault type k. This matrix reflects the number of votes received by various fault types at each location under the action of different feature matchers.

[0108] To achieve a comprehensive decision-making objective, we can use the voting statistics corresponding to fault type k at each scale. Obtain the voting statistics corresponding to fault type k. As shown in equation (14).

[0109] (14)

[0110] In the formula, S represents the total number of scales; and This represents the voting statistic calculated for fault type k across all scales.

[0111] By analyzing the vote statistics By performing analysis, the fault type in the distribution network can be determined. If the voting statistics corresponding to fault type k... If the vote count exceeds a pre-set threshold T, the location of the fault in the distribution network can be determined. The corresponding fault type is k.

[0112] In this embodiment, the fault feature vector at each scale is input into different feature matchers in a pre-built distribution network fault type identification model to match the fault feature vector with the fault type, thereby obtaining the voting statistics corresponding to each fault type. Based on the relationship between the voting statistics corresponding to each fault type and the voting statistics threshold, the fault type corresponding to the fault location in the distribution network is determined. By combining multi-scale feature matching technology to analyze the equipment operation risk and external factors at the fault location in the distribution network, and analyzing and matching the feature expressions of distribution network faults at different scales, the changing characteristics of distribution network faults at different time and spatial scales can be effectively captured, thereby improving the accuracy and robustness of distribution network fault type identification.

[0113] To better understand the above method, the following describes in detail an application embodiment of the method for identifying distribution network fault types in this application.

[0114] Currently, fault type identification in distribution networks is mainly based on information such as current and voltage, or on the operating status of protective switches and circuit breakers. However, traditional methods for identifying distribution network fault types rely on relatively limited information sources, resulting in low accuracy when faced with complex situations.

[0115] This embodiment provides a method for identifying distribution network fault types based on multi-scale feature matching, which can effectively achieve accurate perception and effective risk control of key weak nodes in the distribution network, including the following steps:

[0116] Step 1: Construct a power distribution network fault location model:

[0117] A distribution network fault location model can be trained using a backpropagation neural network model. This model can be applied to fault assessment in distribution networks to achieve accurate fault location.

[0118] The learning process of the backpropagation neural network model is as follows: Figure 3 As shown, it mainly includes two steps: the first step is the forward propagation stage, where information is passed from the hidden layer to the output layer; the second step is to use error negative feedback to adjust various parameters in the network.

[0119] The weight coefficients and thresholds of the hidden and output layers of the backpropagation neural network can be initialized; training samples can be obtained based on the action information samples of each protection device and each circuit breaker in the distribution network; and at least one round of training can be performed based on the training samples.

[0120] During the i-th round of training, the training samples are... The formulas for calculating the input and output values ​​of the m-th neuron in the hidden layer of the backpropagation neural network are shown in Equations (1) and (2).

[0121] (1)

[0122] (2)

[0123] In the formula, This represents the input value of the hidden layer. This represents the output value of the hidden layer. This represents the weight coefficients between the input layer and the hidden layer. This represents the threshold of the hidden layer. represents the transfer function of the hidden layer, and N represents the number of neurons in the input layer.

[0124] The output data of the hidden layer during the i-th training round can be obtained from the output values ​​of all neurons in the hidden layer during the i-th training round.

[0125] The formulas for calculating the input value and output value received by the a-th neuron in the output layer are shown in equations (3) and (4).

[0126] (3)

[0127] (4)

[0128] In the formula, This represents the input value of the output layer. Indicates the output value of the output layer; This represents the weight coefficients between the output layer and the hidden layer. Indicates the threshold of the output layer. This represents the activation function of the output layer, and S represents the number of neurons in the hidden layer.

[0129] The output data of the output layer in the i-th training round can be obtained from the output values ​​of all neurons in the output layer in the i-th training round.

[0130] The output error of the training samples in the i-th round can be obtained from the output data and error function of the output layer of the backpropagation neural network during the i-th round of training.

[0131] The weight coefficients between the output layer and the hidden layer can be adjusted according to the output error of the i-th round of the training samples, as shown in Equations (5) and (6), so as to obtain the weight coefficients between the output layer and the hidden layer in the (i+1)-th round.

[0132] (5)

[0133] (6)

[0134] In the formula, This represents the weight adjustment amount between the output layer and the hidden layer. These represent the weight adjustments between the hidden layer and the input layer, respectively. represents the model's learning rate, i represents the number of training iterations, and M represents the number of neurons in the output layer sample.

[0135] The calculation formulas for optimizing the thresholds of the hidden layer and the output layer are shown in Equations (7) and (8), thus obtaining the thresholds of the hidden layer and the output layer in the (i+1)th round.

[0136] (7)

[0137] (8)

[0138] In the formula, This represents the threshold of the hidden layer. This represents the threshold of the output layer.

[0139] If the output error of the training sample in the i-th round is greater than the preset threshold, then the (i+1)-th round of training is performed; if the output error of the training sample in the i-th round is less than or equal to the preset threshold (iteration termination condition), then training stops, and the distribution network fault location model is obtained based on the weight coefficients of the hidden layer and the output layer in the (i+1)-th round and the threshold in the (i+1)-th round. The preset threshold can be determined according to the actual situation.

[0140] Step 2: Construct a distribution network fault type identification model based on multi-scale feature matching:

[0141] The feature extraction process provided in this embodiment is as follows:

[0142] Assume that the real-time situational awareness data of the fault location in the distribution network is a one-dimensional signal. , for signal The extracted fault feature data is subjected to multi-scale wavelet decomposition to obtain a series of fault feature sub-data at different scales. It can also be called a sub-signal. , where r is used to represent the scale.

[0143] Based on the sub-signals at each scale Calculate the corresponding fault feature vector As shown in equation (12).

[0144] (12)

[0145] In the formula, Represents sub-signal The mean, Represents sub-signal standard deviation Sub-signal The power spectral density function.

[0146] Based on the above feature extraction process, features can be extracted from historical event data of the power distribution network to obtain fault feature vector sample data and normal feature vector sample data.

[0147] A distribution network fault type identification model can be trained based on fault feature vector sample data and normal feature vector sample data. This model can be applied to fault assessment in distribution networks to achieve accurate fault type identification.

[0148] Step 3: Solve the distribution network fault type identification model based on the ensemble learning algorithm:

[0149] To improve the accuracy of distribution network fault type identification, the distribution network fault type identification model can be solved based on ensemble learning algorithms.

[0150] The calculation and analysis of the distribution network fault type identification model are as follows:

[0151] Assume that the distribution network fault type identification model has y feature matchers, labeled M1, M2, ..., My. The fault feature vector at each scale can be input into each feature matcher in the pre-built distribution network fault type identification model. Each feature matcher performs a feature matching operation on the fault feature vector at each scale to match the fault feature vector with the fault type, obtaining the matching result corresponding to the fault feature vector at each scale, thus obtaining the corresponding matching result set for each scale.

[0152] To achieve integrated decision-making, a voting mechanism is introduced. This is applied to the fault location of each distribution network. It can be the position coordinates of the sliding window. The system analyzes the outputs of all feature matchers and makes a final decision based on a voting mechanism. More specifically, it analyzes the matching result set for each scale to obtain the voting statistics for each fault type.

[0153] The voting statistics for fault type k at the s-th level scale As shown in equation (13).

[0154] (13)

[0155] In the formula, This indicates that for the s-th layer scale feature matcher Mb at position The location is determined to be a fault type. Similarity score; This indicates an indicator function. The function's value is 1 when the condition within the parentheses is true, and 0 otherwise.

[0156] Within each scale (also known as scale space), it is possible to utilize Obtain the voting statistics matrix for fault type k. This matrix reflects the number of votes received by various fault types at each location under the action of different feature matchers.

[0157] To achieve a comprehensive decision-making objective, we can use the voting statistics corresponding to fault type k at each scale. Obtain the voting statistics corresponding to fault type k. As shown in equation (14).

[0158] (14)

[0159] In the formula, S represents the total number of scales; and This represents the voting statistic calculated for fault type k across all scales.

[0160] By analyzing the vote statistics By performing analysis, the fault type in the distribution network can be determined. If the voting statistics corresponding to fault type k... If the vote count exceeds a pre-set threshold T, the location of the fault in the distribution network can be determined. The corresponding fault type is k.

[0161] The distribution network fault type identification method provided in this embodiment addresses issues such as false tripping, failure to tripping, and data distortion in switch quantity information. It leverages the powerful adaptive and generalization capabilities of a backpropagation neural network, using the action information of various protection devices and circuit breakers as input and the fault probability of each area of ​​the distribution network as output. By constructing a mapping relationship between input and output, a distribution network fault location model is trained to achieve accurate fault location. Furthermore, based on multi-scale feature matching technology, the fault type corresponding to the fault location in the distribution network is obtained. This allows for the extraction and integration of fault features at multiple scales, thereby improving the accuracy and robustness of fault type identification. The aforementioned multi-scale feature matching technology effectively captures the changing characteristics of distribution network faults at different time and spatial scales. Combined with real-time situational awareness data and historical event data, and considering equipment operation risks and coupled external factors such as meteorology, it can be used to assess, analyze, and predict potential risks in the distribution network.

[0162] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0163] Based on the same inventive concept, this application also provides a distribution network fault type identification device for implementing the above-described distribution network fault type identification method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more distribution network fault type identification device embodiments provided below can be found in the limitations of the distribution network fault type identification method described above, and will not be repeated here.

[0164] In one exemplary embodiment, such as Figure 4 As shown, a device for identifying the type of fault in a power distribution network is provided, wherein:

[0165] The fault feature data acquisition module 401 is used to extract features from real-time situational awareness data, equipment operation risk data and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network.

[0166] Fault feature vector acquisition 402 is used to perform multi-scale wavelet decomposition on the fault feature data to obtain the fault feature vector at each scale.

[0167] The fault type identification module 403 is used to input the fault feature vector into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

[0168] In one embodiment, the device further includes a model training module for: training a distribution network fault location model based on a backpropagation neural network model; inputting the action information of each protection device and each circuit breaker in the distribution network into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network, so as to determine the fault location of the distribution network.

[0169] In one embodiment, the model training module is further configured to: initialize the weight coefficients and thresholds of the hidden layer and output layer of the backpropagation neural network; obtain training samples based on the action information samples of each protection device and each circuit breaker in the distribution network; perform at least one round of training, wherein, in each round of training, the training samples are input into the backpropagation neural network to obtain the current round input data and current round output data of the hidden layer and output layer of the backpropagation neural network; obtain the current round output error of the training samples based on the current round output data of the output layer of the backpropagation neural network; update the next round weight coefficients and next round thresholds of the hidden layer and output layer based on the current round output error of the training samples; if the current round output error of the training samples is greater than a preset threshold, then proceed to the next round of training; if the current round output error of the training samples is less than or equal to the preset threshold, then stop training, and obtain the distribution network fault location model based on the next round weight coefficients and next round thresholds of the hidden layer and output layer.

[0170] In one embodiment, the fault feature data acquisition module 401 is further configured to: preprocess the real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data; extract the mean and standard deviation of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the time-frequency features corresponding to the fault location in the distribution network; extract the spectral density values ​​of the preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the frequency domain features corresponding to the fault location in the distribution network; and obtain the fault feature data corresponding to the fault location in the distribution network based on the time-frequency features and frequency domain features corresponding to the fault location in the distribution network.

[0171] In one embodiment, the fault feature vector acquisition 402 is further used to: perform multi-scale wavelet decomposition on the fault feature data to obtain fault feature sub-data at different scales; and obtain the fault feature vector at each scale based on the fault feature sub-data at each scale.

[0172] In one embodiment, the fault type identification module 403 is further configured to: input the fault feature vector at each scale into different feature matchers in a pre-built distribution network fault type identification model to match the fault feature vector with the fault type, thereby obtaining a matching result set for each scale; statistically analyze the matching result set for each scale to obtain the voting statistics for each fault type; if the voting statistics for one fault type exceed the voting statistics threshold, then the fault type is determined as the fault type corresponding to the fault location in the distribution network.

[0173] Each module in the aforementioned power distribution network fault type identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0174] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data for embodiments of a method for identifying distribution network fault types. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for identifying distribution network fault types.

[0175] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0176] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0177] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0178] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0179] 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, data stored, data displayed, 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 the relevant data must comply with relevant regulations.

[0180] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0181] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0182] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for identifying fault types in a distribution network, characterized in that, The method includes: Feature extraction is performed on real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network. The fault feature data is subjected to multi-scale wavelet decomposition to obtain the fault feature vector at each scale. The fault feature vector is input into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

2. The method according to claim 1, characterized in that, Before extracting features from real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network, the method further includes: A fault location model for the distribution network is trained based on the backpropagation neural network model. The operation information of each protection device and each circuit breaker in the distribution network is input into the distribution network fault location model to obtain the fault probability of each protection device and each circuit breaker in the distribution network, so as to determine the fault location of the distribution network.

3. The method according to claim 2, characterized in that, The method of training a distribution network fault location model based on a backpropagation neural network model includes: Initialize the weight coefficients and thresholds of the hidden and output layers of the backpropagation neural network; Training samples are obtained based on the action information samples of each protection device and each circuit breaker in the distribution network. At least one round of training is performed, wherein in each round of training, the training samples are input into the backpropagation neural network to obtain the input data and output data of the hidden layer and output layer of the backpropagation neural network for the current round; based on the output data of the output layer of the backpropagation neural network for the current round, the output error of the training samples for the current round is obtained; based on the output error of the training samples for the current round, the weight coefficients and thresholds of the hidden layer and output layer for the next round are updated. If the current output error of the training sample is greater than a preset threshold, then the next round of training is performed; if the current output error of the training sample is less than or equal to the preset threshold, then training is stopped, and the distribution network fault location model is obtained based on the weight coefficients and thresholds of the hidden layer and the output layer for the next round.

4. The method according to claim 1, characterized in that, The real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network are used to extract features to obtain fault feature data corresponding to the fault location in the distribution network, including: Real-time situational awareness data, equipment operation risk data, and meteorological data of the fault location in the distribution network are preprocessed to obtain preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data. The mean and standard deviation of preprocessed real-time situational awareness data, equipment operation risk data and meteorological data are extracted to obtain the time-frequency characteristics corresponding to the fault location of the distribution network. Extract the spectral density values ​​of preprocessed real-time situational awareness data, equipment operation risk data, and meteorological data to obtain the frequency domain characteristics corresponding to the fault location of the distribution network; Based on the time-frequency and frequency-domain characteristics corresponding to the fault location in the distribution network, fault characteristic data corresponding to the fault location in the distribution network are obtained.

5. The method according to claim 1, characterized in that, The step of performing multi-scale wavelet decomposition on the fault feature data to obtain fault feature vectors at each scale includes: The fault feature data is subjected to multi-scale wavelet decomposition to obtain fault feature sub-data at different scales; Based on the fault feature sub-data at each scale, the fault feature vector at each scale is obtained.

6. The method according to claim 1, characterized in that, The step of inputting the fault feature vector into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network includes: The fault feature vector at each scale is input into different feature matchers in the pre-built distribution network fault type identification model to match the fault feature vector with the fault type, thereby obtaining the corresponding matching result set for each scale. By statistically analyzing the matching result set for each scale, the voting statistics for each fault type are obtained. If the voting statistics corresponding to one of the fault types exceed the voting statistics threshold, then the fault type is determined as the fault type corresponding to the fault location of the distribution network.

7. A device for identifying the type of fault in a power distribution network, characterized in that, The device includes: The fault feature data acquisition module is used to extract features from real-time situational awareness data, equipment operation risk data and meteorological data of the fault location in the distribution network to obtain fault feature data corresponding to the fault location in the distribution network. Fault feature vector acquisition is used to perform multi-scale wavelet decomposition on the fault feature data to obtain the fault feature vector at each scale. The fault type identification module is used to input the fault feature vector into a pre-built distribution network fault type identification model to obtain the fault type corresponding to the fault location in the distribution network.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images