CN-122026502-A - Time sequence analysis method, system, equipment and medium for distributed energy access power distribution network

CN122026502ACN 122026502 ACN122026502 ACN 122026502ACN-122026502-A

Abstract

The invention discloses a time sequence analysis method, a system, equipment and a medium for a distributed energy access power distribution network, wherein the method comprises the steps of obtaining running data and climate data of a distributed power supply, establishing a photovoltaic output correction model, and calculating to obtain photovoltaic output power; the method comprises the steps of obtaining fan output data, establishing a fan output correction model to calculate and obtain fan output power, establishing a space-time joint probability distribution model based on photovoltaic output power and fan output power, obtaining a central scene and corresponding weight through the space-time joint probability distribution model, establishing and solving an active and reactive power balance equation of a power grid based on the central scene, obtaining node voltage data and establishing a voltage probability distribution function. According to the method, the influence of salt fog and typhoon on the output of the distributed energy sources can be accurately described, the time sequence analysis precision of the power distribution network is improved through efficient scene dimension reduction and probability modeling, the online calculation requirement is met, the voltage risk can be quantized, and the support is provided for safe and stable operation of high-proportion distributed energy sources connected to the power distribution network.

Inventors

ZHANG NA
GUO HUIFANG
ZHENG DING
WU CHUNJI
Lin Yunxuan
LIN SHIZHE
DONG YUNZHOU
CAO LU
Zhu Mingsi
XING BOXIANG
QI MINGLU

Assignees

海南电网有限责任公司信息通信分公司

Dates

Publication Date: 20260512
Application Date: 20251017

Claims (10)

1. The time sequence analysis method for the distributed energy access power distribution network is characterized by comprising the following steps of: Acquiring distributed power supply operation data and climate data, establishing a photovoltaic output correction model through an LSTM neural network, and calculating to acquire photovoltaic output power; Acquiring fan output data, establishing a fan output correction model by combining a power curve shape coefficient, and calculating to acquire fan output power; Based on the photovoltaic output power and the fan output power, constructing a space-time joint probability distribution model, generating a distributed power source output scene set containing correlation through the space-time joint probability distribution model, and preprocessing the distributed power source output scene set to obtain a center scene and corresponding weights; Based on the central scene, the load data and the power grid node admittance matrix, constructing and solving a power grid active and reactive power balance equation to obtain node voltage data, and establishing a voltage probability distribution function by aggregating the node voltage data.
2. The method for timing analysis of a distributed energy access power distribution network of claim 1, wherein calculating the photovoltaic output power comprises: Acquiring exposure time, salt spray concentration sequence, temperature and humidity data of a photovoltaic panel, and inputting the data into a pre-trained LSTM neural network to obtain a first attenuation coefficient; Processing the salt spray concentration sequence data to obtain weighted average salt spray concentration in the history preset time, and obtaining a second attenuation coefficient according to the weighted average salt spray concentration and the corrosion rate formula; and obtaining a target attenuation coefficient according to the first attenuation coefficient and the second attenuation coefficient, establishing a photovoltaic output correction model by combining solar irradiance parameters, and calculating to obtain photovoltaic output.
3. The method for timing analysis of a distributed energy access power distribution network of claim 2, wherein calculating the power output of the wind turbine comprises: Typhoon path forecast data at different moments are obtained, wind speed, turbulence intensity, wind direction mutation frequency and rainfall intensity data are determined according to the typhoon path forecast data, a wind speed and power response model is built based on the wind speed data, and a power curve shape coefficient is obtained; calculating corresponding damping factors according to the turbulence intensity, the wind direction mutation frequency and the rainfall intensity respectively, and fusing to obtain typhoon turbulence damping factors; and combining the shape coefficient of the power curve and the typhoon turbulence damping factor, establishing a fan output correction model, and calculating to obtain the fan output power.
4. A method of time series analysis of a distributed energy access power distribution network as claimed in claim 3 wherein constructing a spatio-temporal joint probability distribution model based on the photovoltaic output power and the fan output power comprises: obtaining a photovoltaic output power generation photovoltaic historical output time sequence at a plurality of moments and a wind power generation wind power historical output time sequence at a plurality of moments; Determining marginal probability distribution functions corresponding to the photovoltaic and wind power historical output time sequences by adopting a nuclear density estimation method, and carrying out probability integral transformation on the photovoltaic and wind power historical output time sequences based on the marginal probability distribution functions to obtain a uniform distribution data matrix; And calculating correlation coefficients among all power supplies based on the uniformly distributed data matrix, constructing a rattan Copula structure, selecting an optimal Copula function for each binary variable pair in the rattan Copula structure, estimating parameters, and generating a space-time joint probability distribution model of photovoltaic and wind power output.
5. The method for time sequence analysis of a distributed energy access power distribution network according to claim 4, wherein generating a distributed power output scene set containing correlation by a space-time joint probability distribution model, preprocessing the distributed power output scene set to obtain a central scene and a corresponding weight, comprises: sampling from the space-time joint probability distribution model, and transforming the sampling result back to the original output space through the inverse function of the marginal probability distribution of each power supply to generate a distributed power supply output scene set containing space-time correlation; Calculating average output vectors of all scenes in the distributed power output scene set, and determining weights of all scenes by means of Markov distances by combining mean vectors and covariance matrixes of historical data matrixes constructed by photovoltaic and wind power historical output; After normalizing all scene weights, classifying the scene set into a plurality of scene subsets by adopting a k-means clustering algorithm, selecting the central scene of each subset, and summing the weights of all the scenes in the subset to serve as updated weights of the corresponding central scenes.
6. The method for time series analysis of a distributed energy access power distribution network according to claim 5, wherein constructing and solving a power grid active and reactive power balance equation to obtain node voltage data comprises: selecting each central scene, determining corresponding reactive power according to the analog output power of the distributed power supply in the central scene, simultaneously acquiring historical load data matched with scene time stamps, and combining a target area power grid structure to obtain a power grid node admittance matrix; respectively constructing an active power balance equation and a reactive power balance equation of the power grid according to the distributed power supply power, the load power demand and the node admittance matrix; Initializing voltage amplitude and phase angle under each central scene, iteratively solving an active power balance equation and a reactive power balance equation of the power grid by adopting a Newton-Lapherson method, calculating net injection quantity of active power and reactive power of nodes each time, judging whether deviation between the net injection quantity and the equation is smaller than a preset error or not until iteration converges, and outputting voltage amplitude and phase angle data of each node.
7. The method of time series analysis of a distributed energy access power distribution network of claim 6, wherein aggregating node voltage data establishes a voltage probability distribution function comprising: Collecting voltage amplitude data of each node obtained by calculation in all central scenes, and generating a voltage sample set corresponding to each node; Based on the dirac function, a voltage probability distribution function is established for each node by combining the central scene weight corresponding to each voltage value in the sample set, the occurrence probability of different voltage values under each scene is obtained through the voltage probability distribution function, and voltage quality assessment and risk management are performed according to the occurrence probability.
8. A time sequence analysis system for a distributed energy access distribution network, applying the method according to any one of claims 1-7, comprising: The photovoltaic data acquisition module is used for acquiring running data and climate data of the distributed power supply, establishing a photovoltaic output correction model through an LSTM neural network and calculating and acquiring photovoltaic output power; the fan data acquisition module is used for acquiring fan output data, establishing a fan output correction model by combining the shape coefficient of the power curve, and calculating and acquiring fan output power; The space-time joint calculation module is used for constructing a space-time joint probability distribution model based on the photovoltaic output power and the fan output power, generating a distributed power output scene set containing correlation through the space-time joint probability distribution model, and preprocessing the distributed power output scene set to obtain a central scene and corresponding weights; The data solving and analyzing module is used for constructing and solving an active and reactive power balance equation of the power grid based on the central scene, the load data and the power grid node admittance matrix to obtain node voltage data, and establishing a voltage probability distribution function by aggregating the node voltage data.
9. An electronic device, comprising: The method comprises the steps of the method for analyzing the time sequence of the distributed energy access distribution network according to any one of claims 1 to 7, wherein the method comprises the steps of storing computer executable instructions, and the processor is used for executing the computer executable instructions.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor perform the steps of the method of time sequence analysis of a distributed energy access distribution network according to any one of claims 1 to 7.

Description

Time sequence analysis method, system, equipment and medium for distributed energy access power distribution network Technical Field The invention relates to the technical field of power distribution networks, in particular to a time sequence analysis method, a system, equipment and a medium for a distributed energy access power distribution network. Background With the advancement of dual carbon targets, distributed power supplies are rapidly evolving. The total photovoltaic overall size of more than 670 regions in the whole country is approximately 170GW. However, the photovoltaic output has strong randomness, intermittence and strong space-time correlation, and the continuous access of the low-voltage side high-proportion renewable energy sources (such as photovoltaic and wind power) to the power distribution network brings great challenges to the safe operation of the power grid, aggravates the three-phase imbalance degree of the power distribution network, and causes the problems of power dumping, unbalanced feeder load, frequent voltage fluctuation out-of-limit and the like. Probability flow (Probabilistic Power Flow, PPF) computation is a key tool to analyze system uncertainty. However, the conventional deterministic power flow analysis can only provide a single operation state, cannot quantitatively evaluate the risk brought by the fluctuation of the distributed energy, and is difficult to meet the requirements of planning and operation of the novel power distribution system. In addition, the existing probability power flow calculation method still has significant defects when facing high-proportion distributed energy access, firstly, most researches assume that wind and light output is subject to independent distribution or simple correlation, complex space-time coupling characteristics influenced by microclimate (such as typhoon and salt fog) cannot be accurately described, so that generated scene distortion is caused, secondly, the traditional method adopts Monte Carlo simulation, power flow calculation is required to be carried out for more than ten thousands times, the calculation efficiency is low, and the online analysis requirement is difficult to meet. Therefore, a time sequence trend analysis method capable of accurately describing the space-time correlation of distributed energy sources, achieving high calculation efficiency and accurately evaluating the operation risk of the high-proportion distributed energy sources accessing the lower power distribution network is needed. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a time sequence analysis method for a distributed energy access power distribution network, which solves the problems that the traditional method has low modeling precision and poor calculation efficiency and is difficult to accurately evaluate the running risk of the power grid when the high-proportion distributed energy is accessed into the power distribution network. In order to solve the technical problems, the invention provides the following technical scheme: in a first aspect, the present invention provides a method for analyzing a time sequence of a distributed energy access power distribution network, including: Acquiring distributed power supply operation data and climate data, establishing a photovoltaic output correction model through an LSTM neural network, and calculating to acquire photovoltaic output power; Acquiring fan output data, establishing a fan output correction model by combining a power curve shape coefficient, and calculating to acquire fan output power; Based on the photovoltaic output power and the fan output power, constructing a space-time joint probability distribution model, generating a distributed power source output scene set containing correlation through the space-time joint probability distribution model, and preprocessing the distributed power source output scene set to obtain a center scene and corresponding weights; Based on the central scene, the load data and the power grid node admittance matrix, constructing and solving a power grid active and reactive power balance equation to obtain node voltage data, and establishing a voltage probability distribution function by aggregating the node voltage data. As a preferable scheme of the time sequence analysis method of the distributed energy access distribution network, the method comprises the following steps of calculating and obtaining photovoltaic output power: Acquiring exposure time, salt spray concentration sequence, temperature and humidity data of a photovoltaic panel, and inputting the data into a pre-trained LSTM neural network to obtain a first attenuation coefficient; Processing the salt spray concentration sequence data to obtain weighted average salt spray concentration in the history preset time, and obtaining a second attenuation coefficient according to