CN-121983999-A - Power distribution network fault self-healing method considering distributed power supply characteristics and related device

CN121983999ACN 121983999 ACN121983999 ACN 121983999ACN-121983999-A

Abstract

A power distribution network fault self-healing method considering the characteristic of a distributed power supply and a related device are provided, wherein the method comprises the steps of collecting historical output data of the distributed new energy, and obtaining a typical output scene through clustering; the method comprises the steps of respectively establishing an output prediction model based on wavelet transformation and LSTM aiming at each typical scene, matching the typical scene according to the current real-time output data, obtaining output data in the future T time by utilizing the corresponding prediction model, constructing a line fault self-healing model containing indexes of running severity, load loss degree and economy of a power distribution network, extracting the maximum and minimum values in the future output data as extreme output when the line is in fault, substituting the maximum and minimum values into the self-healing model to solve a non-fault line reconstruction path, and controlling the switching action according to the path to realize self-healing. According to the invention, through scene clustering and extreme output prediction, the influence of the randomness and fluctuation of the output of the distributed new energy on the self-healing process is effectively reduced, and the running stability and the power supply reliability of the system after the self-healing of the fault are improved.

Inventors

FENG WANLI
LIU BINGHAO
Yan Fangbin
LI LIE
XIONG PING
SU LEI
CAI XUAN
CAO KAN
WEI MINGJIANG
ZHU XIAOQIAN
Chen Ziya
Gu Ruoshi
ZHANG TIANYUN

Assignees

国网湖北省电力有限公司电力科学研究院

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. The utility model provides a distribution network fault self-healing method considering distributed power supply characteristics, which is characterized by comprising the following steps: step S1, collecting historical output data of distributed new energy sources of a power distribution network, and classifying the historical output data by using a clustering method to obtain typical output scenes of the distributed new energy sources; Step S2, respectively establishing a distributed new energy output prediction model corresponding to each typical output scene according to the typical output scene; Step S3, collecting real-time output data of the distributed new energy source in the current period, determining a typical output scene to which the real-time output data belongs, inputting the real-time output data into a distributed new energy source output prediction model of a corresponding scene, and obtaining output data of the distributed new energy source in future T time; And when the power distribution network has a line fault, extracting maximum output data and minimum output data from the output data in the future T time obtained in the step S3 as extreme output data, substituting the extreme output data into the line fault self-healing model to solve, obtaining a reconstructed path of a non-fault line, and controlling a switch to act according to the reconstructed path to realize the self-healing of the line fault of the power distribution network.
2. The method for self-healing a power distribution network fault taking distributed power characteristics into consideration as set forth in claim 1, wherein in step S1, the historical output data is classified by using a clustering method to obtain a typical output scenario of a distributed new energy, and the method specifically includes: Selecting k samples from the historical output data X randomly as initial clustering centers, and calculating the distance between the rest samples in the X and the initial clustering centers; after division is completed each time, the average data of each cluster is used for updating the cluster center, iteration is carried out until the cluster center is stable, k cluster results are output and used as k typical output scenes X k .
3. The method for self-healing a power distribution network fault taking distributed power characteristics into consideration as set forth in claim 1, wherein in step S2, a distributed new energy output prediction model corresponding to each typical output scene is established, and specifically includes: Processing each typical output scene X k by utilizing wavelet transformation, and decomposing to obtain time sequence data W Xk of different frequency bands: (2); Wherein a 0 represents a scale parameter, τ 0 represents a translation parameter, and m and n are the scale of the signal and the position of a sampling point respectively; respectively inputting time sequence data of each frequency band into an LSTM model for training to obtain predicted data L Xk of the corresponding frequency band; And coupling and reconstructing the predicted data L Xk of each frequency band to form a distributed new energy output prediction model in the typical output scene.
4. The power distribution network fault self-healing method considering distributed power characteristics according to claim 1, wherein the objective function of the line fault self-healing model in step S4 is: (6); Wherein R is a power distribution network operation severity index, F is a load loss index, E is an economic index, delta 1 、δ 2 and delta 3 respectively represent the weight of the power distribution network operation severity index R, the load loss index F and the economic index E; the calculation formula of the operation severity index R of the power distribution network is as follows: (3); Wherein U i represents the voltage of a node i, R ij ,X ij represents the resistance and reactance parameters of a line respectively, P ij ,Q ij represents the active power and reactive power of a node j connected with the node i respectively, and N represents the node number of the power distribution network; the calculation formula of the load loss degree index F is as follows: (4); Wherein C p,t represents the electricity price at time T, λ l represents an importance index of the load, P l,t represents the load lost at time T, T represents the predicted time length, Δt represents the single period length at the time of calculation; The calculation formula of the economic index E is as follows: (5); S represents the switching operation cost, H i,t and H i,t-1 respectively represent the switching states of the interconnection switches i in the self-healing process, and M represents the number of the interconnection switches of the power distribution network; the constraints of the line fault self-healing model comprise power flow constraints, power supply output constraints, node voltage constraints, branch current constraints and radial network constraints.
5. A power distribution network fault self-healing device considering distributed power characteristics, comprising: The data acquisition module is used for acquiring historical output data and real-time output data of the distributed new energy sources in the power distribution network; the clustering module is used for carrying out clustering analysis on the historical output data to obtain a typical output scene of the distributed new energy; the prediction model construction module is used for respectively constructing a distributed new energy output prediction model corresponding to each typical output scene according to the typical output scene; The scene recognition and prediction module is used for recognizing the typical output scene to which the real-time output data belongs, calling an output prediction model of the corresponding scene, and predicting to obtain output data of the distributed new energy source in the future T time; The self-healing decision module is internally provided with a line fault self-healing model formed by a power distribution network operation severity index, a load loss index and an economic index, and is used for extracting maximum output data and minimum output data from output data in the future T time as extreme output data when the power distribution network has a line fault, substituting the extreme output data into the line fault self-healing model to solve, and obtaining a reconstruction path of a non-fault line; and the control execution module is used for controlling corresponding switching actions according to the reconstruction path to realize the self-healing of the power distribution network line faults.
6. The power distribution network fault self-healing apparatus taking into account distributed power characteristics according to claim 5, wherein the prediction model construction module includes: the wavelet decomposition unit is used for carrying out wavelet transformation on the output data of each typical output scene to obtain time sequence data of different frequency bands; The LSTM training unit is used for respectively inputting the time sequence data of each frequency band into the LSTM model for training to obtain the predicted data of each frequency band; And the reconstruction unit is used for coupling and reconstructing the prediction data of each frequency band to form an output prediction model of the corresponding scene.
7. The power distribution network fault self-healing apparatus taking into account distributed power characteristics according to claim 5, wherein the objective function of the line fault self-healing model is: (6); Wherein R is a power distribution network operation severity index, F is a load loss index, E is an economic index, delta 1 、δ 2 and delta 3 respectively represent the weight of the power distribution network operation severity index R, the load loss index F and the economic index E; the calculation formula of the operation severity index R of the power distribution network is as follows: (3); Wherein U i represents the voltage of a node i, R ij ,X ij represents the resistance and reactance parameters of a line respectively, P ij ,Q ij represents the active power and reactive power of a node j connected with the node i respectively, and N represents the node number of the power distribution network; the calculation formula of the load loss degree index F is as follows: (4); Wherein C p,t represents the electricity price at time T, λ l represents an importance index of the load, P l,t represents the load lost at time T, T represents the predicted time length, Δt represents the single period length at the time of calculation; The calculation formula of the economic index E is as follows: (5); Wherein S represents the switch action cost, H i,t and H i,t-1 respectively represent the switch states of the interconnection switch i in the self-healing process, and M represents the number of the interconnection switches of the power distribution network.
8. The power distribution network fault self-healing system taking into account distributed power characteristics according to claim 5, wherein the constraints of the line fault self-healing model include power flow constraints, power supply output constraints, node voltage constraints, branch current constraints, and radial network constraints.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 4 when the computer program is executed.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any one of claims 1 to 4.

Description

Power distribution network fault self-healing method considering distributed power supply characteristics and related device Technical Field The invention relates to the field of power distribution network fault self-healing, in particular to a power distribution network fault self-healing method considering the characteristics of a distributed power supply and a related device. Background The power distribution network has close relation with the productivity of enterprises and the life quality of residents, and the guarantee of stable and reliable power supply is a primary task. The power distribution network adopts a power supply mode of open loop design and closed loop operation, and under a fault state, the quick recovery power supply of a non-fault line power supply area is realized by adjusting the on-off state of a line switch, so that the core target of the self-healing of the power distribution network is realized. However, with the intermittent and fluctuating power supply caused by the generation of distributed new energy, it becomes more difficult to make a power distribution network self-heal rapidly under abnormal and faulty operation conditions. The traditional mathematical programming method based on deterministic description and various intelligent optimization algorithms based on heuristic method are difficult to effectively solve the problems, and the traditional method does not consider fluctuation of distributed new energy output, and the obtained optimal self-healing operation solution can influence the stability of subsequent operation due to uncertainty of distributed new energy output, so that the operation efficiency and the power supply capacity of the power distribution network are reduced. Disclosure of Invention The invention provides a power distribution network fault self-healing method and a related device considering the characteristics of a distributed power supply, which can reduce the influence of uncertainty of the output of the distributed new energy on the power distribution network fault self-healing and improve the operation reliability of the power distribution network. In order to achieve the technical purpose, the invention adopts the following technical scheme: A power distribution network fault self-healing method considering the characteristics of a distributed power supply comprises the following steps: step S1, collecting historical output data of distributed new energy sources of a power distribution network, and classifying the historical output data by using a clustering method to obtain typical output scenes of the distributed new energy sources; Step S2, respectively establishing a distributed new energy output prediction model corresponding to each typical output scene according to the typical output scene; Step S3, collecting real-time output data of the distributed new energy source in the current period, determining a typical output scene to which the real-time output data belongs, inputting the real-time output data into a distributed new energy source output prediction model of a corresponding scene, and obtaining output data of the distributed new energy source in future T time; And when the power distribution network has a line fault, extracting maximum output data and minimum output data from the output data in the future T time obtained in the step S3 as extreme output data, substituting the extreme output data into the line fault self-healing model to solve, obtaining a reconstructed path of a non-fault line, and controlling a switch to act according to the reconstructed path to realize the self-healing of the line fault of the power distribution network. Further, in step S1, the step of classifying the historical output data by using a clustering method to obtain a typical output scenario of the distributed new energy source specifically includes: Selecting k samples from the historical output data X randomly as initial clustering centers, and calculating the distance between the rest samples in the X and the initial clustering centers; after division is completed each time, the average data of each cluster is used for updating the cluster center, iteration is carried out until the cluster center is stable, k cluster results are output and used as k typical output scenes X k. Further, in step S2, the building of the distributed new energy output prediction model corresponding to each typical output scene specifically includes: Processing each typical output scene X k by utilizing wavelet transformation, and decomposing to obtain time sequence data W Xk of different frequency bands: (2); Wherein a 0 represents a scale parameter, τ 0 represents a translation parameter, and m and n are the scale of the signal and the position of a sampling point respectively; respectively inputting time sequence data of each frequency band into an LSTM model for training to obtain predicted data L Xk of the corresponding frequency band; And coupling and recons