Method and device for determining fault reason information of power distribution network and readable storage medium

Abstract

The invention discloses a method and device for determining fault reason information of a power distribution network and a readable storage medium. The method comprises the following steps: acquiring an initial fault signal of the power distribution network under the condition that the power distribution network has a fault; preprocessing the initial fault signal to obtain a target fault signal; and inputting the target fault signal into a target recognition model for recognition to obtain fault reason information of the power distribution network, the target recognition model being obtained by training a fault signal sample of the power distribution network, and the fault reason information being used for representing the reason of the fault of the power distribution network. The technical problem of low fault cause determination efficiency of the power distribution network is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault monitoring, and in particular to a method, device and readable storage medium for determining fault cause information of a distribution network. Background Art

[0002] At present, new energy sources such as photovoltaic power generation and fuel cells are connected to the distribution network in the form of distributed power sources (Distributed Generation, referred to as DG) through diversified power electronic devices, resulting in the "double high" characteristics of a high proportion of renewable energy and a high proportion of power electronic devices in the distribution network. Accurate identification of the cause of distribution network failure is the main method to speed up line repair, reduce power outage time and quickly restore power supply. It is also a key technology to further improve the automation level and power supply reliability of the new energy distributed access distribution network.

[0003] In the related technologies, the distribution network has a complex structure, with a mixture of overhead lines and cables, numerous branch lines, and a harsh environment, which is prone to various types of faults. The method for identifying the cause of single-phase grounding faults in the distribution network mainly relies on "manual line patrol", in which operators patrol along the line to find the cause of the fault, which not only consumes a lot of manpower and material resources, prolongs the power outage time, but also easily leads to misjudgment. Therefore, there is a technical problem of low efficiency in determining the cause of the fault in the distribution network.

[0004] With respect to the above-mentioned technical problem of low efficiency in determining the cause of faults in the distribution network, no effective solution has been proposed so far. Summary of the invention

[0005] Embodiments of the present invention provide a method, device and readable storage medium for determining fault cause information of a distribution network, so as to at least solve the technical problem of low efficiency in determining the fault cause of the distribution network.

[0006] According to one aspect of an embodiment of the present invention, a method for determining fault cause information of a distribution network is provided. The method may include: when a fault occurs in the distribution network, obtaining an initial fault signal of the distribution network; preprocessing the initial fault signal to obtain a target fault signal; inputting the target fault signal into a target recognition model for recognition to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

[0007] Optionally, preprocessing the initial fault signal to obtain the target fault signal includes: extracting a valid fault signal from the initial fault signal; and preprocessing the valid fault signal to obtain the target fault signal.

[0008] Optionally, preprocessing the effective fault signal to obtain a target fault signal includes: decomposing the effective fault signal into multiple effective fault sub-signals; performing noise reduction processing on the multiple effective fault sub-signals to obtain a noise-reduced effective fault signal; and slicing the noise-reduced effective fault signal to obtain a target fault signal.

[0009] Optionally, slicing the effective fault signal after noise reduction to obtain a target fault signal includes: obtaining slicing parameters, wherein the slicing parameters are used to indicate a rule for slicing the effective fault signal; slicing the effective fault signal after noise reduction according to the slicing parameters to obtain the sliced ​​effective fault signal; and performing feature extraction on the sliced ​​effective fault signal to obtain the target fault signal.

[0010] Optionally, performing feature extraction on the sliced ​​effective fault signal to obtain the target fault signal includes: extracting multidimensional features from the sliced ​​effective fault signal; and encoding the multidimensional features to obtain the target fault signal.

[0011] Optionally, the method for determining the fault cause information of the distribution network further includes: using the fault signal samples to train an initial recognition model to obtain a target recognition model.

[0012] According to another aspect of an embodiment of the present invention, a device for determining fault cause information of a distribution network is also provided. The device may include: an acquisition unit, used to acquire an initial fault signal of the distribution network when a fault occurs in the distribution network; a preprocessing unit, used to preprocess the initial fault signal to obtain a target fault signal; and an identification unit, used to input the target fault signal into a target identification model for identification to obtain fault cause information of the distribution network, wherein the target identification model is obtained by training using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

[0013] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method for determining the fault cause information of the distribution network in an embodiment of the present invention.

[0014] According to another aspect of an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein when the program is run, the method for determining fault cause information of a power distribution network in an embodiment of the present invention is executed.

[0015] According to another aspect of the embodiments of the present invention, a computer program product is provided, which includes computer instructions, and when the computer instructions are executed by a processor, the method for determining fault cause information of a power distribution network in the embodiments of the present invention is implemented.

[0016] In an embodiment of the present invention, when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained; the initial fault signal is preprocessed to obtain a target fault signal; the target fault signal is input into a target recognition model for identification to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network. In other words, the embodiment of the present invention processes the initial fault signal of the distribution network, identifies the target fault signal using a target recognition model, and obtains the fault cause information of the distribution network, thereby avoiding the problem of consuming a large amount of human and material resources, solving the technical problem of low efficiency in determining the fault cause of the distribution network, and achieving the technical effect of improving the efficiency in determining the fault cause of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 is a flow chart of a method for determining fault cause information of a distribution network according to an embodiment of the present invention;

[0019] Figure 2 is a flow chart of a method for identifying typical fault causes of a new power distribution system based on HHO-RF according to an embodiment of the present invention;

[0020] Figure 3 It is a schematic diagram of a device for determining fault cause information of a distribution network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, functional component or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, functional components or devices.

[0023] According to an embodiment of the present invention, an embodiment of a method for determining fault cause information of a distribution network is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0024] Figure 1 is a flow chart of a method for determining fault cause information of a distribution network according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:

[0025] Step S101: when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained.

[0026] In the technical solution provided in the above step S101 of the present invention, the initial fault signal may also be referred to as a distribution network grounding fault signal.

[0027] In this embodiment, when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained. For example, the initial fault signal of the distribution network is obtained by a collection device. This is only an example and does not limit the specific method of obtaining the initial fault signal of the distribution network.

[0028] For example, a fault traveling wave acquisition device is installed in the branch line of the distribution network, and seven different types of grounding fault conditions are simulated through artificial grounding tests, and distribution network grounding fault signals caused by different reasons are effectively collected.

[0029] Step S102: pre-processing the initial fault signal to obtain a target fault signal.

[0030] In the technical solution provided in the above step S102 of the present invention, the preprocessing may at least include: noise reduction, slicing and multi-dimensional feature extraction and other operations.

[0031] In this embodiment, after the initial fault signal of the distribution network is acquired in step S101, the initial fault signal is preprocessed to obtain a target fault signal. For example, the initial fault signal is denoised by empirical mode decomposition, which is only an illustrative example and does not limit the specific method of preprocessing the initial fault signal.

[0032] For example, the collected data is preprocessed, including: using a signal time window of 0-0.5ms to avoid the impact of the power electronic equipment control system when DG is connected. The frequency window is in the range of 20-100kHz to avoid the impact of a large number of harmonics generated by the nonlinear characteristics of power electronic equipment when DG is connected; using Empirical Mode Decomposition (EMD) to decompose the data into a finite number of Intrinsic Mode Functions (IMFs) and residual signals, using the spectrum screening method to effectively screen the IMFs, and superimposing the screened IMFs to obtain a de-noised signal.

[0033] Step S103: input the target fault signal into the target identification model for identification to obtain the fault cause information of the distribution network.

[0034] In the technical solution provided in the above step S103 of the present invention, the target recognition model is obtained by training using the fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

[0035] In this embodiment, after the target fault signal is obtained in step S102, the target fault signal is input into the target identification model for identification to obtain the fault cause information of the distribution network. The target identification model may also be called a HHO-RF model.

[0036] Optionally, the collected fault signal samples are decomposed into a training set, a validation set and a test set, and the recognition model is trained using the training set.

[0037] For example, 70% of the preprocessed data is extracted as the training set, 10% as the validation set, and the remaining 20% ​​as the test set. To improve the generalization ability of the model, the data is randomly shuffled before the partitioning operation. The hyperparameters of the HHO optimization algorithm are reasonably set. The training set is input into the HHO-RF model for training, and the model is analyzed for accuracy using the validation set after each iteration until the preset number of iterations is reached.

[0038] It should be noted that the above embodiment can be executed by a device for determining fault cause information of a distribution network.

[0039] In the above steps S101 to S103 of the present invention, when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained; the initial fault signal is preprocessed to obtain a target fault signal; the target fault signal is input into a target recognition model for identification to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network. That is to say, the embodiment of the present invention processes the initial fault signal of the distribution network, identifies the target fault signal using a target recognition model, and obtains the fault cause information of the distribution network, thereby avoiding the problem of consuming a large amount of human and material resources, solving the technical problem of low efficiency in determining the fault cause of the distribution network, and achieving the technical effect of improving the efficiency in determining the fault cause of the distribution network.

[0040] The above method of this embodiment is further introduced below.

[0041] As an optional implementation manner, preprocessing the initial fault signal to obtain the target fault signal includes: extracting a valid fault signal from the initial fault signal; and preprocessing the valid fault signal to obtain the target fault signal.

[0042] In this embodiment, an effective fault signal is extracted from an initial fault signal, and the effective fault signal is preprocessed to obtain a target fault signal. For example, the initial fault signal is processed by a signal time window, a frequency window, and a spectrum screening method.

[0043] For example, the signal time window is in the range of 0-0.5ms to avoid the impact of the power electronic equipment control system when DG is connected. The frequency window is in the range of 20-100kHz to avoid the impact of a large number of harmonics generated by the nonlinear characteristics of power electronic equipment when DG is connected.

[0044] As an optional implementation method, the effective fault signal is preprocessed to obtain a target fault signal, including: decomposing the effective fault signal into multiple effective fault sub-signals; performing noise reduction processing on the multiple effective fault sub-signals to obtain a noise-reduced effective fault signal; and slicing the noise-reduced effective fault signal to obtain a target fault signal.

[0045] In this embodiment, the effective fault signal is decomposed into a plurality of effective fault sub-signals, and the plurality of effective fault sub-signals are screened to obtain a noise-reduced effective fault signal. For example, the effective fault signal is decomposed by empirical mode.

[0046] For example, empirical mode decomposition is used to decompose data into a finite number of intrinsic mode functions and residual signals, and the spectral screening method is used to effectively screen the IMFs. The screened IMFs are superimposed to obtain a denoised signal.

[0047] Optionally, after obtaining the effective fault signal after noise reduction, the effective fault signal after noise reduction is sliced ​​to obtain the target fault signal. The specific slicing method is as follows.

[0048] As an optional implementation method, slicing the effective fault signal after noise reduction to obtain a target fault signal includes: obtaining slicing parameters, wherein the slicing parameters are used to indicate a rule for slicing the effective fault signal; slicing the effective fault signal after noise reduction according to the slicing parameters to obtain a sliced ​​effective fault signal; and performing feature extraction on the sliced ​​effective fault signal to obtain a target fault signal.

[0049] In this embodiment, a slicing parameter is obtained, and the effective fault signal after noise reduction is sliced ​​according to the slicing parameter. The slicing parameter may include a slicing step, a working condition data point, etc. The effective fault signal may be called zero-sequence current waveform data.

[0050] For example, the sliding time window method is used for slicing operation to process the zero-sequence current waveform data. Specifically, each operating condition contains N data points. The slicing step is set to x, which means that each sliding will skip x data points for the next slicing. Each sample is set to contain n data points to ensure the sufficiency and representativeness of the sample. After such slicing processing, each operating condition finally contains X data samples, as shown in the following formula (1):

[0051]

[0052] Calculate the variance, power spectrum density, waveform factor and Shannon entropy of the data samples of each working condition, that is, extract the features.

[0053] Optionally, feature extraction is performed on the sliced ​​valid fault signal to obtain a target fault signal. The specific privilege extraction method is shown as follows.

[0054] As an optional implementation manner, extracting features from the sliced ​​effective fault signal to obtain a target fault signal includes: extracting multidimensional features from the sliced ​​effective fault signal; encoding the multidimensional features to obtain the target fault signal.

[0055] In this embodiment, multi-dimensional features are extracted from the sliced ​​effective fault signal, wherein the multi-dimensional features may at least include: variance, power spectrum density, waveform factor and Shannon entropy.

[0056] For example, based on the effective fault signal after slicing, four types of characteristic values, namely variance, power spectrum density, waveform factor and Shannon entropy, are calculated for each working condition data sample.

[0057] Optionally, after extracting the multidimensional features, the multidimensional features are encoded to obtain the target fault signal. For example, Chinese labels in text form are mapped into numerical forms that the model can understand and process, thereby ensuring that the model can effectively learn and recognize these labels.

[0058] As an optional implementation manner, the method for determining the fault cause information of the distribution network further includes: training an initial recognition model using the fault signal sample to obtain a target recognition model.

[0059] In this embodiment, the fault signal samples are input into the initial recognition model for training to obtain the target recognition model, wherein the fault signal samples may also be referred to as a training set.

[0060] Optionally, the initial recognition model is trained using the training set, and the accuracy of the model is analyzed using the validation set after each iteration until a preset number of iterations is reached, thereby ensuring the accuracy of the target recognition model.

[0061] It should be noted that the above embodiment can be executed by a device for determining fault cause information of a distribution network.

[0062] In this embodiment, when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained; the initial fault signal is preprocessed to obtain a target fault signal; the target fault signal is input into a target recognition model for identification to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network. That is to say, the embodiment of the present invention processes the initial fault signal of the distribution network, identifies the target fault signal using a target recognition model, and obtains the fault cause information of the distribution network, thereby avoiding the problem of consuming a large amount of human and material resources, solving the technical problem of low efficiency in determining the fault cause of the distribution network, and achieving the technical effect of improving the efficiency in determining the fault cause of the distribution network.

[0063] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.

[0064] At present, new energy sources such as photovoltaic power generation and fuel cells are connected to the distribution network in a large scale in the form of distributed power sources through power electronic devices, resulting in the "double high" characteristics of a high proportion of renewable energy and a high proportion of power electronic devices in the distribution network. Accurate identification of the cause of distribution network failure is the main method to speed up line repair, reduce power outage time and quickly restore power supply. It is also a key technology to further improve the automation level and power supply reliability of new energy distributed access distribution network.

[0065] In the related art, the distribution network has a complex structure, with a mixture of overhead lines and cables, numerous branch lines, and a harsh environment, which makes it easy for various types of faults to occur. The method for identifying the cause of a single-phase grounding fault in a distribution network mainly relies on "manual line patrol", in which operators patrol along the line to find the cause of the fault, which not only consumes a lot of manpower and material resources and prolongs the power outage time, but also easily leads to misjudgment. Therefore, there is a technical problem of low efficiency in determining the cause of a distribution network fault. At present, no effective solution has been proposed for the above-mentioned technical problem of low efficiency in determining the cause of a distribution network fault.

[0066] However, the embodiment of the present invention proposes a new method for identifying the causes of typical faults in a power distribution system based on HHO-RF. A fault traveling wave acquisition device is installed in the branch line of the distribution network. Seven different types of grounding faults are simulated through artificial grounding tests, and distribution network grounding fault signals caused by different reasons are effectively collected. The data of each working condition collected through the experiment are preprocessed, and 70% of the preprocessed data are extracted as a training set, 10% of the data are used as a validation set, and the remaining 20% ​​of the data are used as a test set. In order to improve the generalization ability of the model, the data is randomly shuffled before the division operation. The hyperparameters of the HHO optimization algorithm are reasonably set. The training set is input into the HHO-RF model for training, and the model is analyzed for accuracy using the validation set after each iteration until the preset number of iterations is reached. The reserved test set is input into the trained HHO-RF model to identify the causes of typical faults in the distribution network, which solves the technical problem of low efficiency in determining the causes of faults in the distribution network, and achieves the technical effect of improving the efficiency of determining the causes of faults in the distribution network.

[0067] The embodiments of the present invention are further described below.

[0068] Figure 2 1 is a flow chart of a method for identifying typical fault causes of a new power distribution system based on HHO-RF according to an embodiment of the present invention. The analysis method comprises the following steps:

[0069] Step S201, collecting a distribution network grounding fault signal.

[0070] In this embodiment, a fault traveling wave acquisition device is installed in the branch line of the distribution network, and seven different types of grounding fault conditions are simulated through an artificial grounding test, and distribution network grounding fault signals caused by different reasons are effectively collected.

[0071] Optionally, obtaining zero-sequence current waveform data under various operating conditions of distribution network grounding faults caused by different reasons also includes analyzing actual grounding faults using real fault recording data, thereby obtaining more real and rich zero-sequence current waveform data.

[0072] Step S202: preprocessing the collected fault signal.

[0073] In this embodiment, a time window of 0-0.5 ms and a frequency window of 20-100 kHz are used for each operating condition data collected through experiments, and then the data is preprocessed by noise reduction, slicing, multi-dimensional feature extraction, and other operations.

[0074] Optionally, since the machine learning model cannot effectively recognize Chinese labels, the signal labels need to be encoded before model training to facilitate subsequent judgment of whether the output results are correct.

[0075] Optionally, the collected data is preprocessed, including: using a signal time window in the range of 0-0.5ms to avoid the influence of the power electronic equipment control system when the DG is connected. The frequency window uses a range of 20-100kHz to avoid the influence of a large number of harmonics generated by the nonlinear characteristics of the power electronic equipment when the DG is connected; using empirical mode decomposition to decompose the data into a finite number of intrinsic mode functions and residual signals, using the spectrum screening method to effectively screen the IMFs, superimposing the screened IMFs to obtain a noise-reduced signal; using the sliding time window method to perform slicing operations to process the zero-sequence current waveform data. Specifically, each operating condition contains N data points, and the slicing step is set to x, which means that each sliding will skip x data points for the next slicing, and each sample is set to contain n data points to ensure the sufficiency and representativeness of the sample. After such slicing processing, each operating condition finally contains X data samples, as shown in the above formula (1), which will not be repeated here.

[0076] Optionally, four types of eigenvalues, namely, variance, power spectral density, waveform factor and Shannon entropy, are calculated for the data samples of each working condition. When processing text data, machine learning models often face the problem of being unable to effectively identify and understand, especially when Chinese labels are involved. This is because machine learning algorithms usually perform calculations based on numerical values, and Chinese labels, as text information, cannot be directly processed by the model. Therefore, in the preprocessing stage before model training, a crucial step is to encode and convert Chinese labels. This encoding process aims to map Chinese labels in text form into numerical forms that the model can understand and process, thereby ensuring that the model can effectively learn and recognize these labels. Table 1 is a label table corresponding to fault conditions. As shown in Table 1, the grounding fault types and data labels corresponding to different conditions are different.

[0077] Table 1 Fault condition corresponding label table

[0078]

[0079] Step S203, dividing the preprocessed data into data sets.

[0080] In this embodiment, 70% of the preprocessed data is extracted as a training set, 10% of the data is extracted as a validation set, and the remaining 20% ​​of the data is extracted as a test set. In order to improve the generalization ability of the model, the data is randomly shuffled before the division operation.

[0081] Optionally, 70% of the preprocessed data is extracted as a training set, 10% of the data is used as a validation set, and the remaining 20% ​​of the data is used as a test set. In order to improve the generalization ability of the model, the data is randomly shuffled before the division operation, so the data of each working condition is unevenly distributed. The training set, validation set, and test set are shown in the following formulas (2) to (4):

[0082] X t =7X*0.7 (2)

[0083] X v =7X*0.1 (3)

[0084] X test =7X*0.2 (4)

[0085] Step S204: training the HHO-RF model.

[0086] In this embodiment, the hyperparameters of the HHO optimization algorithm are reasonably set. The training set is input into the HHO-RF model for training, and the accuracy of the model is analyzed using the validation set after each iteration until the preset number of iterations is reached.

[0087] Optionally, parameters are set for the Harris Hawk optimization algorithm. Table 2 is a HHO parameter setting table, and the setting results are shown in Table 2. The Harris Hawk algorithm mainly consists of three parts: a search phase, a transition phase, and a development phase.

[0088] Table 2 HHO parameter setting table

[0089]

[0090] Optionally, in the global search phase, the Harris hawk randomly updates its position to search for prey. Its mathematical expression is shown in the following formula (5):

[0091]

[0092] Where X(t) is the position of the eagle at the tth iteration; X rand (t) is a random individual; X r (t) is the prey position; X m (t) is the average position; q and r1~r4 are random numbers belonging to [0,1]; G is the number of population.

[0093] Optionally, the transition phase selects between exploration and exploitation behaviors according to the prey escape energy E. Its expression is shown in the following formula (6):

[0094]

[0095] in, is the initial escape energy; T is the maximum number of iterations.

[0096] Optionally, during the development phase, a variety of different predation strategies are designed based on the escape behavior of the prey. This article mainly introduces two classic strategies. When 0.5<|E|≤1.0 and r≥0.5, the prey has sufficient energy and has a probability of escaping. At this time, the Harris Hawk adopts a soft siege strategy to gradually consume the energy of the prey. The expression is shown in the following formula (7):

[0097] X(t+1)=ΔX(t)-E|JX rabbit (t)-X(t)| (7)

[0098] Where ΔX(t)=X rabbit (t)-X(t) is the distance between the optimal individual and the current individual; J is a random number in [0,2].

[0099] Alternatively, when |E|<0.5 and r>0.5, the prey's energy is exhausted. At this time, the Harris hawk adopts a hard siege strategy to quickly dive and catch the prey. The expression is shown in the following formula (8):

[0100] X(t+1)=X rabbit (t)-E|ΔX(t)| (8)

[0101] Alternatively, the HHO algorithm has shown good performance and application prospects in the field of optimization problem solving due to its advantages such as high convergence accuracy and strong global search capability.

[0102] Optionally, the decision tree is an important component of the random forest algorithm. The decision tree can effectively analyze the given sample and output the final judgment result. At present, the commonly used decision tree construction algorithm is C4.5. This algorithm uses the information gain rate as a criterion to select features and constructs a decision tree through recursive decision making. Given a data set D, the information entropy H(D) is expressed as shown in the following formula (9):

[0103]

[0104] Among them, p i is the probability value of each category.

[0105] Optionally, the penalty coefficient H of feature A in dataset D A (D) is as shown in the following formula (10):

[0106]

[0107] Among them, D i is a subset of the dataset D.

[0108] Optionally, the empirical conditional entropy H(D|A) of feature A in data set D is as shown in the following formula (11):

[0109]

[0110] Optionally, the information gain rate of feature A in data set D is as shown in the following formula (12):

[0111]

[0112] Optionally, the RF algorithm process is as follows: Randomly extract samples from the original data set with replacement to generate a model training subset. It should be noted that the number of model training subsets should be less than the number of samples in the original data set. Randomly select N feature attributes and input them into the model training subset to generate a decision tree. Repeat the above steps M times to generate a random forest model containing M decision trees. Input the data set into the decision tree for fault diagnosis, and use the voting method to summarize the output results of the decision tree and output the final diagnosis result.

[0113] Step S205: input the test set into the HHO-RF model to identify typical fault causes of the distribution network.

[0114] In this embodiment, the reserved test set is input into the trained HHO-RF model to identify typical fault causes of the distribution network.

[0115] Optionally, the training set is fed into the HHO-RF model for training, initializing the HHO algorithm with each hawk representing a potential solution, i.e., a set of specific number and depth of decision trees. During initialization, multiple Harris hawk positions are randomly generated, corresponding to different number and depth of decision trees.

[0116] Optionally, for each Harris Hawk (i.e., each group of different decision tree number and depth combinations), train the corresponding RF model and evaluate its performance on the validation set, where accuracy is used as the optimization target. Use the evaluation result as the fitness value of the Harris Hawk. The higher the fitness value, the better the performance of the Harris Hawk (i.e., parameter combination).

[0117] Optionally, the position of the Harris Hawk is updated by a random search strategy based on its predation behavior. In each generation, the Harris Hawk with the highest fitness is selected as the leader, and its position may be the optimal solution currently found. Repeat the above steps until the maximum number of iterations is reached. After the optimization process is completed, the position of the leading Harris Hawk is output, that is, the optimal number and depth combination of decision trees. The RF model is retrained using this set of optimal parameters, and its final performance is evaluated on the test set. At this point, the HHO-RF model training is completed. It should be noted that the model training time and fault identification accuracy belong to two objective functions of different dimensions. Therefore, the time-varying precision coefficient criterion C is adopted. tva Confirm the model weights, C tva The larger the value, the more the result is in line with expectations, as shown in the following formula (13):

[0118] Ctva=tw1+aw2+vw3 (13)

[0119] Among them, a represents the percentage of correct classification in the validation set; t represents the model training time; v represents the variance of the percentage of correct classification in the validation set; w1 and w3 are assigned to -1; and w2 is assigned to 1.

[0120] Optionally, the reserved test set is input into the trained HHO-RF model to identify the typical fault causes of the distribution network. Table 3 is an example table of typical fault cause identification results. As shown in Table 3, the typical fault cause identification results of the proposed method are compared with those of the commonly used model.

[0121] Table 3 Typical fault cause identification results example table

[0122]

[0123] In this embodiment, a fault traveling wave acquisition device is installed in the branch line of the distribution network, and 7 different types of grounding fault conditions are simulated through artificial grounding tests, and distribution network grounding fault signals caused by different reasons are effectively collected. The data of each working condition collected through the experiment are preprocessed, and 70% of the preprocessed data are extracted as a training set, 10% of the data are used as a validation set, and the remaining 20% ​​of the data are used as a test set. In order to improve the generalization ability of the model, the data is randomly shuffled before the division operation. The hyperparameters of the HHO optimization algorithm are reasonably set. The training set is input into the HHO-RF model for training, and the model is analyzed for accuracy using the validation set after each iteration until the preset number of iterations is reached. The reserved test set is input into the trained HHO-RF model to identify the typical fault causes of the distribution network, which solves the technical problem of low efficiency in determining the fault causes of the distribution network, and achieves the technical effect of improving the efficiency of determining the fault causes of the distribution network.

[0124] According to an embodiment of the present invention, a device for determining fault cause information of a distribution network is also provided. It should be noted that the device for determining fault cause information of a distribution network can be used to execute the method for determining fault cause information of a distribution network in the method embodiment.

[0125] Figure 3 Schematic diagram of a device for determining fault cause information of a distribution network according to an embodiment of the present invention. Figure 3 As shown, the device 300 for determining fault cause information of the power distribution network may include: an acquisition unit 301 , a preprocessing unit 302 and an identification unit 303 .

[0126] The acquisition unit 301 is used to acquire an initial fault signal of the distribution network when a fault occurs in the distribution network.

[0127] The preprocessing unit 302 is used to preprocess the initial fault signal to obtain a target fault signal.

[0128] The identification unit 303 is used to input the target fault signal into the target identification model for identification to obtain the fault cause information of the distribution network, wherein the target identification model is trained using the fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

[0129] Optionally, the preprocessing unit 302 may include: an extraction module, configured to extract a valid fault signal from an initial fault signal; and a preprocessing module, configured to preprocess the valid fault signal to obtain a target fault signal.

[0130] Optionally, the preprocessing module may include: a decomposition submodule, used to decompose the effective fault signal into multiple effective fault sub-signals; a screening submodule, used to perform noise reduction processing on the multiple effective fault sub-signals to obtain the effective fault signal after noise reduction; and a slicing submodule, used to slice the effective fault signal after noise reduction to obtain the target fault signal.

[0131] Optionally, the slicing submodule can also be used to obtain slicing parameters, where the slicing parameters are used to indicate the rules for slicing the valid fault signal; according to the slicing parameters, the denoised valid fault signal is sliced ​​to obtain the sliced ​​valid fault signal; and the feature extraction of the sliced ​​valid fault signal is performed to obtain the target fault signal.

[0132] Optionally, the slicing submodule can also be used to extract multi-dimensional features from the sliced ​​effective fault signal; encode the multi-dimensional features to obtain the target fault signal.

[0133] Optionally, the device 300 for determining fault cause information of a distribution network may further include: a training unit, configured to train an initial recognition model using fault signal samples to obtain a target recognition model.

[0134] In this embodiment, when a fault occurs in the distribution network, an initial fault signal of the distribution network is obtained; the initial fault signal is preprocessed to obtain a target fault signal; the target fault signal is input into a target recognition model for identification to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network. That is to say, the embodiment of the present invention processes the initial fault signal of the distribution network, identifies the target fault signal using a target recognition model, and obtains the fault cause information of the distribution network, thereby avoiding the problem of consuming a large amount of human and material resources, solving the technical problem of low efficiency in determining the fault cause of the distribution network, and achieving the technical effect of improving the efficiency in determining the fault cause of the distribution network.

[0135] According to an embodiment of the present invention, there is further provided a computer-readable storage medium, the storage medium comprising a stored program, wherein the program executes the method for determining fault cause information of a power distribution network in an embodiment of the method.

[0136] According to an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein when the program is run, the method for determining fault cause information of a distribution network in the method embodiment is executed.

[0137] According to an embodiment of the present invention, a computer program product is further provided. The computer program product includes computer instructions. When the computer instructions are executed by a processor, the method for determining fault cause information of a power distribution network in the method embodiment is implemented.

[0138] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0139] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0140] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0141] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0142] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0143] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent functional component, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software functional component, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc., which can store program codes.

[0144] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining fault cause information of a distribution network, characterized in that: include: In case of a fault in the distribution network, obtaining an initial fault signal of the distribution network; Preprocessing the initial fault signal to obtain a target fault signal; The target fault signal is input into a target recognition model for recognition to obtain fault cause information of the distribution network, wherein the target recognition model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

2. The method according to claim 1, characterized in that Preprocessing the initial fault signal to obtain a target fault signal includes: Extracting a valid fault signal from the initial fault signal; The effective fault signal is preprocessed to obtain the target fault signal.

3. The method according to claim 2, characterized in that Preprocessing the effective fault signal to obtain the target fault signal includes: Decomposing the effective fault signal into a plurality of effective fault sub-signals; Performing noise reduction processing on the plurality of valid fault sub-signals to obtain the valid fault signal after noise reduction; The effective fault signal after noise reduction is sliced ​​to obtain the target fault signal.

4. The method according to claim 3, characterized in that Slicing the effective fault signal after noise reduction to obtain the target fault signal includes: Acquire a slicing parameter, wherein the slicing parameter is used to indicate a rule for slicing the valid fault signal; Slicing the effective fault signal after noise reduction according to the slicing parameter to obtain the effective fault signal after slicing; Feature extraction is performed on the sliced ​​effective fault signal to obtain the target fault signal.

5. The method according to claim 4, characterized in that Performing feature extraction on the sliced ​​effective fault signal to obtain the target fault signal includes: Extracting multidimensional features from the sliced ​​effective fault signal; The multi-dimensional features are encoded to obtain the target fault signal.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: The initial recognition model is trained using the fault signal samples to obtain the target recognition model.

7. A device for determining fault cause information of a distribution network, characterized in that: include: An acquisition unit, used for acquiring an initial fault signal of the distribution network when a fault occurs in the distribution network; A preprocessing unit, used for preprocessing the initial fault signal to obtain a target fault signal; An identification unit is used to input the target fault signal into a target identification model for identification, so as to obtain fault cause information of the distribution network, wherein the target identification model is trained using fault signal samples of the distribution network, and the fault cause information is used to indicate the cause of the fault in the distribution network.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 6 when running.

10. A computer program product, characterized in that The computer program product comprises computer instructions, which implement the method according to any one of claims 1 to 6 when executed by a processor.

Images