Weak power distribution network power failure risk intelligent early warning method and device

Abstract

The invention belongs to the technical field of power distribution network operation analysis, and particularly relates to a weak power distribution network power failure risk intelligent early warning method and device, and the method comprises the steps: collecting weak power distribution network data of a target sampling period; based on the weak power distribution network data of the target sampling period, utilizing a pre-established risk early warning model to obtain a power failure risk early warning result of the weak power distribution network in the prediction time; the risk early warning model is established by using historical weak power distribution network data. According to the technical scheme provided by the invention, accurate mining and intelligent early warning of the power failure risk of the weak power distribution network are realized, the early warning precision and efficiency are improved, the risk level of the weak power distribution network is effectively reduced, and the power supply reliability of the weak power distribution network is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network operation analysis, and specifically relates to an intelligent early warning method and device for power outage risk in a weak distribution network. Background Art

[0002] The distribution network has a complex structure, a wide range, a large number of equipment types and quantities, and is scattered in location, and is vulnerable to natural disasters. After a sudden disaster, the use of distributed generation (DG) and topology transformation can quickly restore important loads and reduce fault losses, but the fault recovery capacity of the distribution network is limited, and load power outages may still occur. Therefore, in order to adapt to the development trend of the power system, higher requirements are put forward for distribution network risk warning. Effective warning of power outage risks in the operation of the distribution network is a key way to improve the reliability of distribution network power supply.

[0003] The power supply areas at the end of the distribution network feeder, such as rural distribution networks and remote power supply areas, are located in harsh operating environments (such as floods, lightning strikes, wildfires, ice disasters and earthquakes). Due to their long transmission distances, large line impedance, low load density and mismatch between load power consumption and distributed power generation timing, large-scale distributed energy access will lead to difficulties in absorbing distributed resources in the area, over-limit voltage at the end of the feeder, and overall weak grid characteristics, which will further reduce the power supply reliability of the relevant power supply areas.

[0004] Reliable power outage risk warning technology can effectively ensure the power supply reliability of weak distribution networks, while the refined warning of weak distribution networks needs to rely on reasonable warning indicators and rich data sources. At present, the risk warning methods of distribution networks are mainly divided into two categories: mathematical statistics and artificial intelligence algorithms. The mathematical statistics method uses historical fault data as the source and constructs a fault probability prediction model through the quantitative relationship between disaster and fault information. The artificial intelligence algorithm establishes a fault prediction model by introducing disaster information as an input variable.

[0005] However, among the existing methods, the prediction model constructed by mathematical and statistical methods cannot fully reflect the impact of meteorological factors on the failure rate, making it difficult to achieve accurate and reliable early warning; and the artificial intelligence algorithm is subject to disaster impact data. Since there are many factors that affect the distribution network due to disasters, the input variables of the artificial intelligence algorithm cannot accurately reveal the impact of disasters, and the accuracy of the prediction will be greatly reduced. Summary of the invention

[0006] In order to overcome the problems existing in the above-mentioned related technologies, the present invention provides an intelligent early warning method and device for power outage risk of a weak distribution network.

[0007] According to a first aspect of an embodiment of the present invention, there is provided an intelligent early warning method for power outage risk of a weak distribution network, comprising:

[0008] Collect weak distribution network data in the target sampling period;

[0009] Based on the weak distribution network data of the target sampling period, using a pre-established risk warning model, obtaining a power outage risk warning result of the weak distribution network at the predicted time;

[0010] The risk warning model is established using historical weak distribution network data.

[0011] Preferably, the weak distribution network data based on the target sampling period uses a pre-established risk warning model to obtain a power outage risk warning result of the weak distribution network at the predicted time, including:

[0012] Performing data preprocessing on the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period;

[0013] Using the principal component analysis method, feature extraction is performed on the weak distribution network data of the target sampling period after the processing to obtain the target optimal feature information;

[0014] Normalizing the target optimal feature information to obtain normalized target optimal feature information;

[0015] The normalized target optimal feature information is input into the risk warning model, and a power outage risk warning result of the weak distribution network at the predicted time is output.

[0016] Preferably, the process of establishing the risk warning model includes:

[0017] Collect historical weak distribution network data and historical power outage risk warning results;

[0018] Performing data preprocessing on the historical weak distribution network data to obtain processed historical weak distribution network data;

[0019] Using the principal component analysis method to extract features from the processed historical weak distribution network data, and obtaining historical optimal feature information;

[0020] Normalizing the historical optimal feature information to obtain normalized historical optimal feature information;

[0021] Constructing a data set using the normalized historical optimal feature information and the historical power outage risk warning results;

[0022] The improved support vector machine model is trained and verified using the data set to obtain the risk warning model.

[0023] Preferably, the using the data set to train and verify the improved support vector machine model includes:

[0024] Dividing the data set into a training set and a test set; training the improved support vector machine model using the training set to obtain a trained improved support vector machine model;

[0025] The trained improved support vector machine model is verified using the test set. If the verification is successful, the trained improved support vector machine model is the risk warning model. If the verification fails, the hyperparameters of the improved support vector machine model are adjusted, and the improved support vector machine model is retrained until the verification is successful.

[0026] Preferably, the process of establishing the improved support vector machine model includes:

[0027] The penalty factor and kernel function of the support vector machine model are optimized using the multiverse algorithm to obtain the optimal penalty factor and the optimal kernel function;

[0028] The improved support vector machine model is established using the optimal penalty factor and the optimal kernel function.

[0029] Preferably, the method of optimizing the penalty factor and kernel function of the support vector machine model using the multiverse algorithm to obtain the optimal penalty factor and the optimal kernel function includes:

[0030] Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor, and value range of kernel function;

[0031] Randomly initializing the universe position according to the value range of the penalty factor and the value range of the kernel function;

[0032] Calculate the expansion rate of each universe and rank the expansion rates of all universes;

[0033] Based on the ranking of the expansion rates, a roulette wheel mechanism is used to select white holes from all universes;

[0034] Update the wormhole existence probability and travel distance value;

[0035] Based on the updated wormhole existence probability and the updated travel distance value, the universe position and the optimal universe are updated;

[0036] It is determined whether the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, the universe position is reinitialized until the optimal penalty factor and the optimal kernel function are obtained.

[0037] Preferably, the calculation formula for initializing the cosmic position includes:

[0038]

[0039] In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

[0040] Preferably, the calculation formula of the roulette mechanism includes:

[0041]

[0042] In the above formula, x ij is the jth variable of the initialized i-th universe, x kjis the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

[0043] Preferably, the calculation formula for the probability of wormhole existence includes:

[0044]

[0045] The calculation formula of the travel distance value includes:

[0046]

[0047] In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, and p is the accuracy of iterative development.

[0048] Preferably, the calculation formula for updating the universe position and the optimal universe includes:

[0049]

[0050] In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

[0051] Preferably, the kernel function in the improved SVM model is a radial basis kernel function RBF.

[0052] According to a second aspect of an embodiment of the present invention, there is provided an intelligent early warning device for power outage risk of a weak distribution network, comprising:

[0053] A collection unit, used to collect weak distribution network data in a target sampling period;

[0054] A prediction unit, configured to obtain a power outage risk warning result of the weak distribution network at a predicted time based on the weak distribution network data of the target sampling period and using a pre-established risk warning model;

[0055] The risk warning model is established using historical weak distribution network data.

[0056] Preferably, the prediction unit comprises:

[0057] A first processing module, configured to perform data preprocessing on the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period;

[0058] A first feature extraction module is used to extract features from the weak distribution network data of the target sampling period after the processing by using a principal component analysis method to obtain target optimal feature information;

[0059] A first normalization module, used for normalizing the target optimal feature information to obtain normalized target optimal feature information;

[0060] The prediction module is used to input the normalized target optimal feature information into the risk warning model and output the power outage risk warning result of the weak distribution network at the predicted time.

[0061] Preferably, it further comprises: an establishing unit, used to establish the risk warning model; the establishing unit comprises:

[0062] The collection module is used to collect historical weak distribution network data and historical power outage risk warning results;

[0063] A second processing module is used to perform data preprocessing on the historical weak distribution network data to obtain processed historical weak distribution network data;

[0064] A second feature extraction module is used to extract features from the processed historical weak distribution network data using a principal component analysis method to obtain historical optimal feature information;

[0065] A second normalization module, used to perform normalization processing on the historical optimal feature information to obtain normalized historical optimal feature information;

[0066] A construction module, used to construct a data set using the normalized historical optimal feature information and the historical power outage risk warning results;

[0067] The acquisition module is used to train and verify the improved support vector machine model using the data set to obtain the risk warning model.

[0068] Preferably, the acquisition module includes:

[0069] A division submodule is used to divide the data set into a training set and a test set; a training submodule is used to train the improved support vector machine model using the training set to obtain a trained improved support vector machine model;

[0070] The verification submodule is used to verify the trained improved support vector machine model using the test set. If the verification is successful, the trained improved support vector machine model is the risk warning model; if the verification fails, the hyperparameters of the improved support vector machine model are adjusted, and the improved support vector machine model is retrained until the verification is successful.

[0071] Preferably, the establishment unit further includes: an establishment module for establishing the improved support vector machine model; the establishment module includes: a first acquisition submodule for optimizing the penalty factor and kernel function of the support vector machine model using a multiverse algorithm to obtain an optimal penalty factor and an optimal kernel function;

[0072] The second acquisition submodule is used to establish the improved support vector machine model by using the optimal penalty factor and the optimal kernel function.

[0073] Preferably, the first acquisition submodule is specifically used for:

[0074] Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor, and value range of kernel function;

[0075] Randomly initializing the universe position according to the value range of the penalty factor and the value range of the kernel function;

[0076] Calculate the expansion rate of each universe and rank the expansion rates of all universes;

[0077] Based on the ranking of the expansion rates, a roulette wheel mechanism is used to select white holes from all universes;

[0078] Update the wormhole existence probability and travel distance value;

[0079] Based on the updated wormhole existence probability and the updated travel distance value, the universe position and the optimal universe are updated;

[0080] It is determined whether the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, the universe position is reinitialized until the optimal penalty factor and the optimal kernel function are obtained.

[0081] Preferably, the calculation formula for initializing the cosmic position includes:

[0082]

[0083] In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

[0084] Preferably, the calculation formula of the roulette mechanism includes:

[0085]

[0086] In the above formula, x ij is the jth variable of the initialized i-th universe, x kjis the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

[0087] Preferably, the calculation formula for the probability of wormhole existence includes:

[0088]

[0089] The calculation formula of the travel distance value includes:

[0090]

[0091] In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, and p is the accuracy of iterative development.

[0092] Preferably, the calculation formula for updating the universe position and the optimal universe includes:

[0093]

[0094] In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

[0095] Preferably, the kernel function in the improved SVM model is a radial basis kernel function RBF.

[0096] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0097] The memory is used to store one or more programs;

[0098] When the one or more programs are executed by the at least one processor, the intelligent early warning method for power outage risk of a weak distribution network is implemented.

[0099] According to a fourth aspect of an embodiment of the present invention, there is provided a readable storage medium having an execution program stored thereon, and when the execution program is executed, the intelligent early warning method for power outage risk of a weak distribution network is implemented.

[0100] The technical solution provided by the present invention has the following beneficial effects:

[0101] The present invention provides an intelligent early warning method and device for power outage risk of a weak distribution network. The method collects weak distribution network data of a target sampling period, obtains power outage risk early warning results of the weak distribution network at a predicted time based on the weak distribution network data of the target sampling period, and utilizes a pre-established risk early warning model. The method achieves accurate discovery and intelligent early warning of power outage risks of the weak distribution network, improves early warning accuracy and efficiency, effectively reduces the risk level of the weak distribution network, and improves the power supply reliability of the weak distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0103] Figure 1 It is a flow chart of an intelligent early warning method for power outage risk of a weak distribution network provided by an embodiment of the present invention;

[0104] Figure 2 It is a flow chart of an intelligent early warning method for power outage risk of a weak distribution network provided by an embodiment of the present invention;

[0105] Figure 3 It is a structural block diagram of an intelligent early warning device for power outage risk of a weak distribution network provided by an embodiment of the present invention;

[0106] Figure 4 It is a structural block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0107] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the following embodiments are part of the embodiments of the present invention, rather than 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 are within the scope of protection of the present invention.

[0108] Embodiment 1

[0109] The present invention provides an intelligent early warning method for power outage risk of weak distribution network, such as Figure 1 As shown, the following steps are included:

[0110] Step 11: Collect weak distribution network data in the target sampling period;

[0111] Step 12: Based on the weak distribution network data of the target sampling period, using the pre-established risk warning model, obtain the power outage risk warning result of the weak distribution network at the predicted time;

[0112] The risk warning model is established using historical weak distribution network data.

[0113] In some embodiments, weak distribution network data can be collected from multiple sources. For example, weak distribution network data can be collected from distribution automation systems, data acquisition and monitoring control systems, production management systems, user power consumption information collection systems, geographic information systems, and weather information systems; weak distribution network data can include, but are not limited to: historical operation data of weak distribution networks, equipment inventory data, reasons for power outage risks of weak distribution networks, power outage time, number of power outages, and power shortage of weak distribution networks.

[0114] It should be noted that the embodiment of the present invention does not limit the “target sampling period”, and it can be set by those skilled in the art according to engineering needs, experimental data or expert experience.

[0115] Further, step 12 includes:

[0116] Step 121: preprocessing the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period;

[0117] Step 122: extracting features from the processed weak distribution network data of the target sampling period using a principal component analysis method to obtain target optimal feature information;

[0118] Step 123: normalizing the target optimal feature information to obtain normalized target optimal feature information;

[0119] Step 124: Input the normalized target optimal feature information into the risk warning model, and output the power outage risk warning result of the weak distribution network at the predicted time.

[0120] It should be noted that the "data preprocessing", "principal component analysis method" and "normalization processing" methods involved in the embodiments of the present invention are well known to those skilled in the art, and therefore, their specific implementation methods are not described in detail. In some embodiments, data preprocessing is the filling of missing data and the processing of abnormal data, and abnormal data processing includes the presentation of redundant data and the correction of erroneous data.

[0121] The present invention uses the principal component analysis method to identify and extract feature information closely related to the power outage risk of weak distribution networks from a large amount of data, thereby improving the efficiency and accuracy of intelligent early warning of power outage risk of weak distribution networks. The target optimal feature information is normalized to make it suitable for the input of the SVM model (i.e., the risk early warning model).

[0122] In some embodiments, the power outage risk warning result may include, but is not limited to: a power outage risk score and a risk warning level;

[0123] Among them, the power outage risk score can be quantified from 0 to 1, and the corresponding risk levels are as follows: 0-0.3 is low risk, 0.3-0.7 is medium risk, and 0.7-1 is high risk.

[0124] Furthermore, the method further comprises: Step 10: establishing a risk warning model; Step 10 comprises:

[0125] Step 101: Collect historical weak distribution network data and historical power outage risk warning results;

[0126] Step 102: preprocessing the historical weak distribution network data to obtain processed historical weak distribution network data;

[0127] Step 103: extracting features from the processed historical weak distribution network data using a principal component analysis method to obtain historical optimal feature information;

[0128] Step 104: normalizing the historical optimal feature information to obtain normalized historical optimal feature information;

[0129] Step 105: constructing a data set using the normalized historical optimal feature information and historical power outage risk warning results;

[0130] Step 106: Use the data set to train and verify the improved support vector machine model to obtain a risk warning model.

[0131] Specifically, the kernel function in the improved SVM model is a radial basis kernel function RBF.

[0132] Further, step 106 includes:

[0133] Step 1061: divide the data set into a training set and a test set; Step 1062: train the improved support vector machine model using the training set to obtain a trained improved support vector machine model;

[0134] Step 1063: Use the test set to verify the trained improved support vector machine model. If the verification is successful, the trained improved support vector machine model is a risk warning model. If the verification fails, adjust the hyperparameters of the improved support vector machine model and retrain the improved support vector machine model until the verification is successful.

[0135] Further, step 1062 includes:

[0136] The normalized historical optimal feature information in the training set is used as the input layer training sample of the improved support vector machine model, and the historical power outage risk warning results in the training set are used as the output layer training sample of the improved support vector machine model. The improved support vector machine model is trained to obtain the trained improved support vector machine model.

[0137] Further, step 1063 includes:

[0138] The normalized historical optimal feature information in the test set is input into the trained improved support vector machine model to output the predicted power outage risk warning result;

[0139] The prediction accuracy is determined based on the predicted power outage risk warning results and the historical power outage risk warning results in the test set. If the prediction accuracy is greater than or equal to the accuracy threshold, the verification is successful; if the prediction accuracy is less than the accuracy threshold, the verification fails.

[0140] It should be noted that the embodiment of the present invention does not limit the "accuracy threshold", which can be set by those skilled in the art according to engineering needs, experimental data or expert experience.

[0141] Furthermore, step 10 also includes: step 100: establishing an improved support vector machine model; step 100 includes:

[0142] Step 1001: Optimizing the penalty factor and kernel function of the support vector machine model using a multiverse algorithm to obtain an optimal penalty factor and an optimal kernel function;

[0143] Step 1002: Establish an improved support vector machine model using the optimal penalty factor and the optimal kernel function.

[0144] It can be understood that the present invention provides an intelligent early warning method for power outage risk of weak distribution network based on MVO-SVM. Multi-Verse Optimization (MVO) is a relatively novel meta-heuristic algorithm, which is derived from black holes, white holes and wormholes in multiverse theory. As a group optimization algorithm that simulates the expansion and contraction process of the universe, it has powerful global search capabilities, fast search capabilities and high adaptability; Support Vector Machine (SVM) is an important tool in the field of pattern recognition and regression, which is used for prediction of high-dimensional, nonlinear and small sample problems, and can effectively overcome the calculation problems caused by excessive prediction deviation values, local extreme values ​​and too high dimensions of conventional methods. The present invention can effectively improve the parameter optimization ability of SVM by optimizing the penalty factor C and kernel function G (γ) parameters of SVM by using the MVO algorithm, thereby improving the early warning performance of the risk early warning model, and improving the early warning accuracy and efficiency.

[0145] Further, step 1001 includes:

[0146] Step 1001a: Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor and value range of kernel function;

[0147] In some embodiments, the range of the penalty factor may be, but is not limited to, [0.1, 1000]; the range of the kernel function may be, but is not limited to, [0.01, 1000];

[0148] Step 1001b: randomly initialize the universe position according to the value range of the penalty factor and the value range of the kernel function, wherein each universe and individual in the universe represents a set of penalty factors and kernel functions;

[0149] Specifically, the calculation formula for initializing the universe position includes:

[0150]

[0151] In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2jis the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe;

[0152] Step 1001c: Calculate the expansion rate of each universe and sort the expansion rates of all universes;

[0153] Step 1001d: Select white holes from all universes using a roulette wheel mechanism based on the order of expansion rates;

[0154] Specifically, the calculation formula of the roulette mechanism includes:

[0155]

[0156] In the above formula, x ij is the jth variable of the initialized i-th universe, x kj is the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe;

[0157] Step 1001e: Update the wormhole existence probability and travel distance value;

[0158] Specifically, the calculation formula for the probability of wormhole existence includes:

[0159]

[0160] The calculation formula for the travel distance value includes:

[0161]

[0162] In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dris the travel distance value, p is the accuracy of iterative development; in some embodiments, the minimum probability of wormhole existence Wep min =1, the maximum probability of wormhole existence Wep max =0.2, p is 6;

[0163] Step 1001f: Based on the updated wormhole existence probability and the updated travel distance value, update the universe position and the optimal universe;

[0164] Specifically, the calculation formula for updating the universe position and the optimal universe includes:

[0165]

[0166] In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively;

[0167] Step 1001g: Determine whether the expansion rate of each universe after the universe position is updated reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the universe position is updated reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, return to step 1001b.

[0168] The intelligent early warning method for power outage risk of weak distribution network proposed in the present invention aims at the problem of repeated power outage accidents in weak distribution network. Firstly, data and information of various information systems of weak distribution network are collected and optimal features are extracted; secondly, a power outage risk early warning model based on SVM is constructed, and the penalty factor C and kernel function G(γ) parameters of SVM are optimized by MVO algorithm to improve the prediction performance of SVM model; finally, intelligent early warning of power outage risk is carried out based on MVO-SVM fusion model, so as to realize accurate discovery and reliable early warning of power outage risk of weak distribution network, effectively reduce the risk level of weak distribution network, and improve the power supply reliability of weak distribution network.

[0169] To further illustrate the above-mentioned intelligent early warning method for power outage risk of weak distribution network, the present invention provides a specific example, such as Figure 2 As shown, the following steps are included:

[0170] S1. Historical weak distribution network data fusion, including preprocessing of acquired multi-source data;

[0171] In the present invention, data fusion includes two aspects, namely: filling missing data and processing abnormal data, and abnormal data processing includes the presentation of redundant data and the correction of erroneous data;

[0172] The multi-source data sources of weak distribution networks include: distribution automation system, data acquisition and monitoring control system, production management system, user power consumption information collection system, geographic information system, weather information system, etc.;

[0173] Multi-source data of weak distribution networks include: historical operation data of weak distribution networks, equipment inventory data, causes of power outage risks of weak distribution networks, power outage time, number of power outages, and power shortage of weak distribution networks;

[0174] Historical operation data include: grid structure data, voltage value, current value, active / reactive power, equipment status, fault records; equipment ledger data include: data of overhead lines, cable lines, switches, distribution transformers and other distribution equipment; in addition, it also includes: temperature, humidity, wind speed, equipment location data, etc.

[0175] The missing data can be filled by Lagrange interpolation method, the redundant data can be directly eliminated, and the wrong data can be corrected by Lagrange interpolation method;

[0176] S2, feature extraction of historical weak distribution network outage information;

[0177] In order to improve the efficiency and accuracy of intelligent early warning of power outage risk of weak distribution network, it is necessary to first identify and extract feature information closely related to the power outage risk of weak distribution network from a large amount of data. In the present invention, the optimal feature information closely related to the power outage risk of weak distribution network is extracted based on the fusion result of step S1;

[0178] In the present invention, the principal component analysis method can be used to extract the optimal feature information, and the extracted historical optimal feature information is shown in Table 1.

[0179] Table 1 Historical optimal feature information

[0180]

[0181]

[0182] S3. Construction of risk warning model of intelligent early warning model for power outage risk of weak distribution network;

[0183] The present invention completes the intelligent early warning of power outage risk for weak distribution networks by constructing an SVM model. The input of the model is the historical optimal feature information of step S2, and the output is the power outage risk early warning result;

[0184] Based on the historical optimal feature information extracted in step S2, data standardization is performed on it, that is, data normalization, so that it is suitable for the input of the SVM model, and the normalized historical optimal feature information is obtained;

[0185] It is worth noting that the purpose of data normalization is to eliminate the differences in dimensions and magnitudes between different variables to improve the accuracy of the SVM model. In the present invention, the Min-Max normalization method is selected, the data size range is [0,1], and the function expression is:

[0186]

[0187] In the above formula, x' is the normalized historical optimal feature information, x is the original historical optimal feature information, and x max is the maximum value of the original historical optimal feature information, x min is the minimum value of the original historical optimal feature information;

[0188] It is worth noting that the regression performance of the SVM model depends largely on the selection of the penalty factor C and the kernel function G(γ). The radial basis kernel function (RBF) is widely used in many complex classification and regression tasks due to its powerful nonlinear mapping ability and stable performance. Therefore, in the present invention, the kernel function G of the SVM selects RBF.

[0189] Before using the model for intelligent warning of power outage risk, the parameter value range of the penalty factor C and the kernel function G in the model is set as follows: C = [0.1, 1000], G = [0.01, 1000];

[0190] Furthermore, the MVO algorithm can be used to optimize the penalty factor C and kernel function G(γ) parameters of SVM to obtain the optimal C and G parameters. The basic principle of the Multi-Verse Optimization (MVO) algorithm is described below:

[0191] The multiverse algorithm is a relatively new meta-heuristic algorithm, which originates from black holes, white holes and wormholes in the multiverse theory. The basic principle of MVO is to simulate the transfer of matter in the universe from white holes to black holes through wormhole tunnels, so as to achieve a relatively stable state. In this invention, the optimal C and G parameters are found by simulating the expansion and contraction of multiple universes. The algorithm principle is as follows, assuming that:

[0192]

[0193] In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

[0194] By establishing a mathematical model to describe the interaction between white holes and black holes, a roulette wheel mechanism is used for selection. In each iteration, the universe is sorted according to its fitness, and a roulette wheel mechanism is used to select a white hole as the object of attention for the current iteration, as shown in the following formula:

[0195]

[0196] In the above formula, x ij is the jth variable of the initialized i-th universe, x kj is the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

[0197] Regardless of the expansion rate, the universe will change its internal structure to optimize the expansion rate, moving its objects toward the optimal universe at that time. The mechanism of updating the universe's position and finding the optimal individual can be expressed as:

[0198]

[0199] In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

[0200] After iteration, the optimal position is continuously updated, and when a new optimal cosmic black hole position is generated, x is solved. ij After that, X j With x ij If x ij The expansion rate is X j The expansion rate is better, then x ij Will replace X j Become a new black hole individual of the next generation, otherwise, X j It will be preserved for the next generation.

[0201] The expressions for wormhole existence probability and travel distance are as follows:

[0202]

[0203] In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, p is the accuracy of iterative development, where Wep max =1,Wep min =0.2, p=6.

[0204] The specific steps of using the MVO algorithm to optimize the penalty factor C and kernel function G(γ) parameters of SVM are as follows:

[0205] S31: Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor and value range of kernel function;

[0206] S32: According to the value range of the penalty factor and the value range of the kernel function, the universe position is randomly initialized using formula (2);

[0207] S33: Calculate the expansion rate of each universe using formula (3), and sort the expansion rates of all universes, and select white holes from all universes using a roulette mechanism based on the sorting of the expansion rates;

[0208] S34: Update the wormhole existence probability and travel distance value using formulas (5) and (6);

[0209] S35: Based on the updated wormhole existence probability and the updated travel distance value, update the universe position and the optimal universe using formula (4);

[0210] S36: Determine whether the expansion rate of each universe after the universe position is updated reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the universe position is updated reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, return to step S32.

[0211] Furthermore, the improved support vector machine model is established using an optimal penalty factor and an optimal kernel function.

[0212] Furthermore, when using the SVM model for training, the input is divided into a training set and a test set in a ratio of 7:3, where the training set is used for model training and the test set is used to verify the model accuracy to obtain a risk warning model;

[0213] Furthermore, the risk warning model is used to carry out intelligent warning of power outage risk, that is, weak distribution network data of the target sampling period is collected, and the weak distribution network data of the target sampling period is used as the input of the risk warning model, and the intelligent warning result of power outage risk is output;

[0214] Furthermore, the intelligent warning results of power outage risk include power outage risk score and risk warning level, among which the power outage risk score can be quantified from 0 to 1, and the corresponding risk levels are as follows: 0-0.3 is low risk, 0.3-0.7 is medium risk, and 0.7-1 is high risk.

[0215] The invention provides an intelligent early warning method for power outage risk of weak distribution network based on MVO-SVM, which can effectively improve the accuracy of power outage risk warning of weak distribution network and improve the power supply reliability of weak distribution network;

[0216] The present invention uses a multiverse algorithm to optimize the C and G parameters of SVM, outputs the optimal C and G parameters, and then constructs an intelligent early warning model for power outage risk to perform intelligent early warning of power outage risk, thereby improving the early warning performance of the model;

[0217] By constructing power outage risk scores and level indicators, the present invention can more intuitively display the intelligent early warning results of power outage risks and provide guidance for on-site operation and maintenance personnel;

[0218] The present invention uses a power outage risk warning model to perform intelligent power outage risk warning, which can filter power outage information more quickly.

[0219] While improving the speed of power outage risk warning, it also enables accurate discovery and reliable warning of power outage risks in weak distribution networks.

[0220] Embodiment 2

[0221] The present invention also provides an intelligent early warning device for power outage risk of weak distribution network, such as Figure 3 As shown, including:

[0222] A collection unit, used to collect weak distribution network data in a target sampling period;

[0223] A prediction unit is used to obtain a power outage risk warning result of the weak distribution network at a predicted time based on the weak distribution network data of a target sampling period and using a pre-established risk warning model;

[0224] The risk warning model is established using historical weak distribution network data.

[0225] Furthermore, the prediction unit comprises:

[0226] A first processing module is used to perform data preprocessing on the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period;

[0227] The first feature extraction module is used to extract features from the weak distribution network data of the processed target sampling period using a principal component analysis method to obtain target optimal feature information;

[0228] A first normalization module, used to perform normalization processing on the target optimal feature information to obtain normalized target optimal feature information;

[0229] The prediction module is used to input the normalized target optimal feature information into the risk warning model and output the power outage risk warning result of the weak distribution network at the predicted time.

[0230] Furthermore, it also includes: establishing a unit for establishing a risk warning model; the establishing unit includes:

[0231] The collection module is used to collect historical weak distribution network data and historical power outage risk warning results;

[0232] The second processing module is used to perform data preprocessing on the historical weak distribution network data to obtain processed historical weak distribution network data;

[0233] The second feature extraction module is used to extract features from the processed historical weak distribution network data using a principal component analysis method to obtain historical optimal feature information;

[0234] The second normalization module is used to normalize the historical optimal feature information to obtain the normalized historical optimal feature information;

[0235] A construction module, used to construct a data set using normalized historical optimal feature information and historical power outage risk warning results;

[0236] The acquisition module is used to train and verify the improved support vector machine model using the data set to obtain a risk warning model.

[0237] Further, the acquisition module includes:

[0238] The partitioning submodule is used to divide the data set into a training set and a test set; the training submodule is used to train the improved support vector machine model using the training set to obtain the trained improved support vector machine model;

[0239] The verification submodule is used to verify the trained improved support vector machine model using the test set. If the verification is successful, the trained improved support vector machine model is a risk warning model; if the verification fails, the hyperparameters of the improved support vector machine model are adjusted, and the improved support vector machine model is retrained until the verification is successful.

[0240] Furthermore, the training submodule is specifically used for:

[0241] The normalized historical optimal feature information in the training set is used as the input layer training sample of the improved support vector machine model, and the historical power outage risk warning results in the training set are used as the output layer training sample of the improved support vector machine model. The improved support vector machine model is trained to obtain the trained improved support vector machine model.

[0242] Furthermore, the verification submodule is used to:

[0243] The normalized historical optimal feature information in the test set is input into the trained improved support vector machine model to output the predicted power outage risk warning result;

[0244] The prediction accuracy is determined based on the predicted power outage risk warning results and the historical power outage risk warning results in the test set. If the prediction accuracy is greater than or equal to the accuracy threshold, the verification is successful; if the prediction accuracy is less than the accuracy threshold, the verification fails.

[0245] Further, the establishment unit further includes: an establishment module for establishing an improved support vector machine model; the establishment module includes: a first acquisition submodule for optimizing the penalty factor and kernel function of the support vector machine model using a multiverse algorithm to obtain an optimal penalty factor and an optimal kernel function;

[0246] The second acquisition submodule is used to establish an improved support vector machine model using an optimal penalty factor and an optimal kernel function.

[0247] Furthermore, the first acquisition submodule is specifically used for:

[0248] Set parameters, including: maximum number of iterations, number of universes, value range of penalty factor and value range of kernel function;

[0249] According to the value range of the penalty factor and the value range of the kernel function, the universe position is randomly initialized;

[0250] Calculate the expansion rate of each universe and rank the expansion rates of all universes;

[0251] Based on the order of expansion rates, white holes are selected from all universes using a roulette wheel mechanism;

[0252] Update the wormhole existence probability and travel distance value;

[0253] Based on the updated wormhole existence probability and the updated travel distance value, the universe position and the optimal universe are updated;

[0254] It is determined whether the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, the universe position is reinitialized until the optimal penalty factor and the optimal kernel function are obtained.

[0255] Furthermore, the calculation formula for initializing the universe position includes:

[0256]

[0257] In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1dis the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

[0258] Furthermore, the calculation formula of the roulette mechanism includes:

[0259]

[0260] In the above formula, x ij is the jth variable of the initialized i-th universe, x kj is the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

[0261] Furthermore, the calculation formula for the probability of wormhole existence includes:

[0262]

[0263] The calculation formula for the travel distance value includes:

[0264]

[0265] In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, and p is the accuracy of iterative development.

[0266] Furthermore, the calculation formulas for updating the universe position and the optimal universe include:

[0267]

[0268] In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

[0269] Furthermore, the kernel function in the improved SVM model is a radial basis kernel function RBF.

[0270] It can be understood that the device embodiment provided above corresponds to the method embodiment described above, and the corresponding specific contents can be referenced to each other and will not be repeated here.

[0271] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0272] Embodiment 3

[0273] like Figure 4 As shown, the present invention also provides an electronic device, which may be a computer device, a single-chip device, an intelligent mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, the processor, and the transceiver component are connected via a bus; the memory may be used to store an execution program, and an exemplary execution program may include instructions; the processor is used to execute the instructions stored in the memory. The memory may also be used to store data, which may be called and / or modified when the instructions are executed.

[0274] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of an intelligent early warning method for power outage risk of a weak distribution network in the above-mentioned embodiment.

[0275] Embodiment 4

[0276] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory), which is a memory device in an electronic device for storing programs and data. It can be understood that the storage medium here can include both built-in storage media in electronic devices and, of course, extended storage media supported by electronic devices. The storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium, which can implement the steps of an intelligent early warning method for power outage risk of a weak distribution network in the above embodiment.

[0277] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0278] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0279] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0280] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0281] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An intelligent early warning method for power outage risk in a weak distribution network, characterized in that: include: Collect weak distribution network data in the target sampling period; Based on the weak distribution network data of the target sampling period, using a pre-established risk warning model, obtaining a power outage risk warning result of the weak distribution network at the predicted time; The risk warning model is established using historical weak distribution network data.

2. The method according to claim 1, characterized in that: The weak distribution network data based on the target sampling period, using a pre-established risk warning model, obtains a power outage risk warning result of the weak distribution network at a predicted time, including: Performing data preprocessing on the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period; Using the principal component analysis method, feature extraction is performed on the weak distribution network data of the target sampling period after the processing to obtain the target optimal feature information; Normalizing the target optimal feature information to obtain normalized target optimal feature information; The normalized target optimal feature information is input into the risk warning model, and a power outage risk warning result of the weak distribution network at the predicted time is output.

3. The method according to claim 1, characterized in that: The process of establishing the risk early warning model includes: Collect historical weak distribution network data and historical power outage risk warning results; Performing data preprocessing on the historical weak distribution network data to obtain processed historical weak distribution network data; Using the principal component analysis method to extract features from the processed historical weak distribution network data, and obtaining historical optimal feature information; Normalizing the historical optimal feature information to obtain normalized historical optimal feature information; Constructing a data set using the normalized historical optimal feature information and the historical power outage risk warning results; The improved support vector machine model is trained and verified using the data set to obtain the risk warning model.

4. The method according to claim 3, characterized in that The step of training and verifying the improved support vector machine model using the data set includes: Dividing the data set into a training set and a test set; training the improved support vector machine model using the training set to obtain a trained improved support vector machine model; The trained improved support vector machine model is verified using the test set. If the verification is successful, the trained improved support vector machine model is the risk warning model. If the verification fails, the hyperparameters of the improved support vector machine model are adjusted, and the improved support vector machine model is retrained until the verification is successful.

5. The method according to claim 3, characterized in that: The process of establishing the improved support vector machine model includes: The penalty factor and kernel function of the support vector machine model are optimized using the multiverse algorithm to obtain the optimal penalty factor and the optimal kernel function; The improved support vector machine model is established using the optimal penalty factor and the optimal kernel function.

6. The method according to claim 5, characterized in that The method of optimizing the penalty factor and kernel function of the support vector machine model by using the multiverse algorithm to obtain the optimal penalty factor and the optimal kernel function includes: Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor, and value range of kernel function; Randomly initializing the universe position according to the value range of the penalty factor and the value range of the kernel function; Calculate the expansion rate of each universe and rank the expansion rates of all universes; Based on the ranking of the expansion rates, a roulette wheel mechanism is used to select white holes from all universes; Update the wormhole existence probability and travel distance value; Based on the updated wormhole existence probability and the updated travel distance value, the universe position and the optimal universe are updated; It is determined whether the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, the universe position is reinitialized until the optimal penalty factor and the optimal kernel function are obtained.

7. The method according to claim 6, characterized in that The calculation formula for initializing the universe position includes: In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

8. The method according to claim 6, characterized in that The calculation formula of the roulette mechanism includes: In the above formula, x ij is the jth variable of the initialized i-th universe, x kj is the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

9. The method according to claim 6, characterized in that The calculation formula for the probability of the wormhole existing includes: The calculation formula of the travel distance value includes: In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, and p is the accuracy of iterative development.

10. The method according to claim 6, characterized in that The calculation formula for updating the universe position and the optimal universe includes: In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

11. The method according to claim 3, characterized in that The kernel function in the improved SVM model is a radial basis kernel function RBF.

12. An intelligent early warning device for power outage risk in a weak distribution network, characterized in that: include: A collection unit, used to collect weak distribution network data in a target sampling period; A prediction unit, configured to obtain a power outage risk warning result of the weak distribution network at a predicted time based on the weak distribution network data of the target sampling period and using a pre-established risk warning model; The risk warning model is established using historical weak distribution network data.

13. The device according to claim 12, characterized in that The prediction unit comprises: A first processing module, configured to perform data preprocessing on the weak distribution network data of the target sampling period to obtain processed weak distribution network data of the target sampling period; A first feature extraction module is used to extract features from the weak distribution network data of the target sampling period after the processing by using a principal component analysis method to obtain target optimal feature information; A first normalization module, used for normalizing the target optimal feature information to obtain normalized target optimal feature information; The prediction module is used to input the normalized target optimal feature information into the risk warning model and output the power outage risk warning result of the weak distribution network at the predicted time.

14. The device according to claim 12, characterized in that The invention also includes: an establishing unit, which is used to establish the risk early warning model; the establishing unit includes: The collection module is used to collect historical weak distribution network data and historical power outage risk warning results; A second processing module is used to perform data preprocessing on the historical weak distribution network data to obtain processed historical weak distribution network data; A second feature extraction module is used to extract features from the processed historical weak distribution network data using a principal component analysis method to obtain historical optimal feature information; A second normalization module, used to perform normalization processing on the historical optimal feature information to obtain normalized historical optimal feature information; A construction module, used to construct a data set using the normalized historical optimal feature information and the historical power outage risk warning results; The acquisition module is used to train and verify the improved support vector machine model using the data set to obtain the risk warning model.

15. The device according to claim 14, characterized in that The acquisition module comprises: A division submodule is used to divide the data set into a training set and a test set; a training submodule is used to train the improved support vector machine model using the training set to obtain a trained improved support vector machine model; The verification submodule is used to verify the trained improved support vector machine model using the test set. If the verification is successful, the trained improved support vector machine model is the risk warning model; if the verification fails, the hyperparameters of the improved support vector machine model are adjusted, and the improved support vector machine model is retrained until the verification is successful.

16. The device according to claim 14, characterized in that The establishment unit further includes: an establishment module for establishing the improved support vector machine model; the establishment module includes: a first acquisition submodule for optimizing the penalty factor and kernel function of the support vector machine model using a multiverse algorithm to obtain an optimal penalty factor and an optimal kernel function; The second acquisition submodule is used to establish the improved support vector machine model by using the optimal penalty factor and the optimal kernel function.

17. The device according to claim 16, characterized in that The first acquisition submodule is specifically used for: Setting parameters, including: maximum number of iterations, number of universes, value range of penalty factor, and value range of kernel function; Randomly initializing the universe position according to the value range of the penalty factor and the value range of the kernel function; Calculate the expansion rate of each universe and rank the expansion rates of all universes; Based on the ranking of the expansion rates, a roulette wheel mechanism is used to select white holes from all universes; Update the wormhole existence probability and travel distance value; Based on the updated wormhole existence probability and the updated travel distance value, the universe position and the optimal universe are updated; It is determined whether the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or whether the current number of iterations reaches the maximum number of iterations. If the expansion rate of each universe after the updated universe position reaches the expansion rate threshold or the current number of iterations reaches the maximum number of iterations, the penalty factor corresponding to the updated optimal universe is the optimal penalty factor, and the kernel function corresponding to the updated optimal universe is the optimal kernel function; otherwise, the universe position is reinitialized until the optimal penalty factor and the optimal kernel function are obtained.

18. The device according to claim 17, characterized in that The calculation formula for initializing the universe position includes: In the above formula, i∈[1,n], n is the total number of universes; j∈[1,d], d is the total number of variables in the universe; Ux is the initialization position of the universe, x 11 is the first variable of the first universe initialized, x 12 is the second variable of the first initialized universe, x 1j is the jth variable of the first universe initialized, x 1d is the dth variable of the first universe initialized, x 21 is the first variable of the second initialized universe, x 22 is the second variable of the second universe initialized, x 2j is the jth variable of the second initialized universe, x 2d is the dth variable of the second initialized universe, x i1 is the first variable of the initialized i-th universe, x i2 is the second variable of the initialized i-th universe, x ij is the jth variable of the initialized i-th universe, x id is the dth variable of the initialized i-th universe, x n1 is the first variable of the initialized nth universe, x n2 is the second variable of the initialized nth universe, x nj is the jth variable of the initialized nth universe, x nd is the dth variable of the initialized nth universe.

19. The device according to claim 17, characterized in that The calculation formula of the roulette mechanism includes: In the above formula, x ij is the jth variable of the initialized i-th universe, x kj is the jth variable of the kth universe selected according to the roulette mechanism, r1 is a random number between 0 and 1, U i is the position of the ith universe, NI(U i ) is the expansion rate of the ith universe.

20. The device according to claim 17, characterized in that The calculation formula for the probability of the wormhole existing includes: The calculation formula of the travel distance value includes: In the above formula, Wep is the probability of wormhole existence, L is the maximum iteration round, l is the current iteration round, Wep min is the minimum probability of wormhole existence, Wep max is the maximum probability of the existence of a wormhole, T dr is the travel distance value, and p is the accuracy of iterative development.

21. The device according to claim 17, characterized in that The calculation formula for updating the universe position and the optimal universe includes: In the above formula, x' ij is the jth variable of the updated i-th universe, x ij is the jth variable of the initialized i-th universe, r2, r3 and r4 are all random numbers between 0 and 1, X j is the updated optimal universe, Wep is the probability of wormhole existence, T dr is the travel distance value, b u,j is the upper limit of the j-th variable, b l,j are the lower limits of the j-th variable respectively.

22. The device according to claim 14, characterized in that The kernel function in the improved SVM model is a radial basis kernel function RBF.

23. An electronic device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the intelligent early warning method for power outage risk of a weak distribution network as described in any one of claims 1 to 11 is implemented.

24. A readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, the intelligent early warning method for power outage risk of a weak distribution network as described in any one of claims 1 to 11 is implemented.

Images