CN-121997696-A - Fault line selection algorithm, system and storage medium for multi-power electronic source power supply system

CN121997696ACN 121997696 ACN121997696 ACN 121997696ACN-121997696-A

Abstract

The invention provides a multi-power electronic source power supply system fault line selection algorithm, a system and a storage medium, wherein the multi-power electronic source power supply system fault line selection algorithm comprises the following steps of 1, establishing a fault history data set; step 2, data preprocessing and feature extraction are carried out, step 3, a fault data training sample is formed, model training is carried out through a random forest algorithm, step 4, data when actual faults occur are obtained, feature extraction is carried out, step 5, a fault line selection model established through the random forest algorithm is compared with a training model through actual measurement data, and then the fault position is determined. The multi-power electronic source power supply system fault line selection algorithm, the system and the storage medium directly perform model training by using a simulation model or fault recording data, extract and analyze characteristic values by using measured data, can realize quick positioning, and can ensure the accuracy of fault line selection of the multi-power source power supply system.

Inventors

WU LEI
LI FENGMING
ZHANG WEIJIN
YIN FENGJIE
XU YINGJIE
WANG ZHENJIE
WU LEILEI

Assignees

中国石油化工集团有限公司
中国石化集团胜利石油管理局有限公司

Dates

Publication Date: 20260508
Application Date: 20241108

Claims (12)

1. The fault line selection algorithm of the multi-power electronic source power supply system is characterized by comprising the following steps of: step 1, establishing a fault history data set; Step 2, data preprocessing and feature extraction are carried out; Step 3, forming a fault data training sample, and performing model training through a random forest algorithm; Step 4, acquiring data when an actual fault occurs, and extracting characteristic features; And 5, comparing the fault line selection model established by using the random forest algorithm with the training model through measured data, and further determining the fault position.
2. The fault line selection algorithm of the multi-power electronic source power supply system according to claim 1, wherein in step 1, a fault history data set of a single-phase ground short circuit, a two-phase inter-phase short circuit and a three-phase short circuit is obtained, and the fault history data set is obtained from a simulation platform or is collected from a fault wave recording device of an actual power station.
3. The fault line selection algorithm for the multi-power electronic source power supply system according to claim 1, wherein in step 2, after the fault history data set is obtained, the data is preprocessed, and non-valid data caused by random noise and interference are filtered.
4. The multiple power electron source supply system fault line selection algorithm according to claim 1, wherein in step 2, the feature extraction comprises three parts of empirical mode decomposition, hilbert spectrum analysis and feature value calculation.
5. The multiple power electron source power supply system fault line selection algorithm according to claim 4, wherein in step 2, the feature extraction specifically comprises: (a) Performing empirical mode decomposition on the zero sequence current data to obtain a mode IMF 1 -IMF n ; (b) Selecting m modes, wherein m is less than or equal to n, and performing Hilbert spectrum analysis on the modes to obtain components of each mode (C) Calculating the mean, standard deviation and energy of the data obtained in the step (b); (d) Features of 3×m×3 dimensions are obtained.
6. The multiple power electron source power supply system fault line selection algorithm according to claim 5, wherein in step c, the formula for calculating the Mean, standard deviation Std, energy is: x i- represents the ith discrete data point, N represents the number of discrete data, x represents the average value, and x (t) represents the real-time data.
7. The multi-power electronic source power supply system fault line selection algorithm according to claim 1, wherein in step 3, features extracted from a fault data set are used as a sample set X, corresponding fault positions are used as a tag set y, training sets x_train and y_train and test sets x_test and y_test are selected in a layered sampling mode, the x_train and the y_train are input into a random forest algorithm RF model for training, a trained fault line selection model is obtained, the x_test and the y_test are input into the trained fault line selection model, and accuracy of the model is evaluated.
8. The fault line selection algorithm for a multi-power electronic source power supply system according to claim 1, wherein in step 4, real-time fault data of the power supply system is collected from a fault recorder of an actual power station and used as a data source of fault line selection.
9. The multiple power electron source power supply system fault line selection algorithm according to claim 8, wherein in step 4, the feature extraction specifically comprises: (a) Empirical mode decomposition is carried out on zero sequence current data in real-time fault data of a power supply system acquired from a fault wave recording device of an actual power station, so as to obtain a mode IMF 1 -IMF n ; (b) Selecting m modes, wherein m is less than or equal to n, and performing Hilbert spectrum analysis on the modes to obtain components of each mode (C) Calculating the mean, standard deviation and energy of the data obtained in the step (b); (d) Features of 3×m×3 dimensions are obtained.
10. The fault line selection algorithm of the multi-power electronic source power supply system according to claim 1, wherein in step 5, the result of feature extraction of the data when the actual fault occurs is compared with the training model structure to determine the fault location, that is, the data of the real-time fault data after feature extraction is added as input into the training random forest algorithm RF model, and the fault line selection location in step 4 is obtained according to the result of the random forest algorithm RF training model.
11. A multiple power electron source power supply system fault line selection system, characterized in that the multiple power electron source power supply system fault line selection system adopts the multiple power electron source power supply system fault line selection algorithm of any one of claims 1-10 to determine a line fault location.
12. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored computer program, wherein the computer program, when running, controls a device in which the computer readable storage medium is located to execute a multiple power electron source power supply system fault line selection algorithm according to any of claims 1-10.

Description

Fault line selection algorithm, system and storage medium for multi-power electronic source power supply system Technical Field The invention relates to the technical field of new energy power generation in electrical engineering, in particular to a fault line selection algorithm, a system and a storage medium of a multi-power electronic source power supply system. Background The photovoltaic power station is used as a common renewable energy power generation system, has the characteristics of cleanness and sustainability, and plays an important role in energy transformation and carbon emission reduction. Along with the continuous expansion of the system scale of the photovoltaic power station, a power supply system with multiple power electronic sources, which is composed of equipment such as a photovoltaic inverter, a transformer, a bus and the like, is formed, the equipment fails to cause the problems of abnormal current, voltage fluctuation, electric energy loss and the like, and the power generation capacity of the photovoltaic power station and the power supply quality of a power grid are reduced. Aiming at the faults, timely and accurate line selection is a key for ensuring the normal operation of the power station. In the chinese patent application with application number cn2015133364. X, a single-phase fault line selection method for a small current grounding system is related, which performs wavelet decomposition on zero sequence current of a feeder line to obtain wavelet energy thereof, and can establish a sample set according to the obtained wavelet energy, normalize the sample set to obtain a normalized sample set X. The SRM neuron model is adopted to initially construct the Spiking neural network, the SpikeProp method is adopted to calculate the weights of the neurons in the input layer H, the neurons in the hidden layer I and the synapses between the neurons in the hidden layer I and the neurons in the output layer J, the selected line Spiking neural network is obtained after the connection weight is corrected, fault data are input into the selected line Spiking neural network, a line marked by phi 0 in the output layer is determined to be a fault line, the operation is convenient, whether the power grid has a ground fault or not can be rapidly and effectively judged, and the line of the ground fault can be judged after the ground fault is determined, so that a basis is provided for timely fault elimination. In the Chinese patent application with the application number of CN201811375953.0, a distribution network single-phase disconnection fault line selection method is related, and comprises the steps of analyzing a topological structure of a distribution network system, collecting voltage and current at a line outlet, calculating a positive sequence current mutation value of a line, comparing the positive sequence current mutation value with a positive sequence current map variable valve, calculating a positive sequence voltage mutation value of the line, comparing the positive sequence voltage mutation value with a positive sequence voltage mutation value, calculating line impedance, comparing the line impedance with a line impedance valve value, obtaining a fault path line, and finally selecting a line with the lowest level in the fault path line as a fault line. The method comprises the steps of determining the composition of mixed measurement data based on a system measurement vector of an active power distribution network, analyzing an associated coefficient matrix by adopting a signal correlation theory, determining a reference measurement time of the mixed measurement system, constructing a delay error function, constructing an active power distribution network linear state estimation model based on a real part and an imaginary part of three-phase node voltage, solving the state quantity of the active power distribution network by adopting an iterative weighted least square method, constructing a fault positioning model by introducing virtual nodes, and determining the position of a fault point. The existing fault line selection algorithm is mainly developed for a low-voltage distribution network in a traditional alternating current power transmission and distribution system, has low applicability to a multi-power electronic source power supply system accessed by a distributed photovoltaic power station, and cannot quickly and accurately realize fault line selection. The prior art is greatly different from the invention, and the related patents mainly relate to insulation monitoring of low-voltage equipment and measurement of a current signal based on an insulation fault loop current sensor, but not relate to fault line selection of a multi-power electronic source power supply system and further relate to an intelligent processing algorithm suitable for fault line selection. The existing fault line selection algorithm is mainly developed for a low-voltage distribution network in a tra