Reactive power optimization method for distribution network based on random matrix and intelligent scene matching

Abstract

A reactive power optimization method for a distribution network based on a random matrix and intelligent scene matching belongs to the technical field of operation control of an AC distribution network. This method directly uses the load data generated during the operation of the distribution network, photovoltaic power generation, electric vehicle charging data, and even environmental data such as temperature, light, and wind speed to construct random matrices respectively; according to the single-loop limit theorem, the characteristics of each random matrix are calculated. Root distribution law, extract statistical features such as average spectral radius, maximum / minimum spectral radius, distribution ratio of characteristic roots outside the ring / on the ring / in the ring; use the principal component analysis method to integrate the above statistical features to obtain the principal component representation The operating state of the distribution network; match the scene characteristics according to the obtained principal components, and use the control strategy in the matched scene as the reactive power optimization control strategy for the current period. The technical scheme of the invention realizes the data-driven reactive power optimization and voltage management of the distribution network, and provides a new way for the optimal operation of the distribution network.

Description

technical field [0001] The invention belongs to the technical field of AC distribution network operation control, and in particular relates to a reactive power optimization method for a distribution network based on random matrix and intelligent scene matching. Background technique [0002] The reactive power optimization and voltage management of the distribution network is an important task for the optimal operation of the distribution network. Small active network loss and node voltage offset make the distribution network operate in a more optimal state. The traditional reactive power optimization method relies on the model and parameters of the distribution network, and repeatedly calculates the power flow of the distribution network during the optimization process. The calculation workload is large, the decision-making time is long, and the adaptability is poor; The artificial intelligence optimization algorithm represented by the neural network, with the increase of t...

AI Technical Summary

Problems solved by technology

However, as the penetration rate of distributed power sources, electric vehicles, and flexible loads gradually increases, the scale of the distribution network continues to expand and its structure becomes increasingly complex. This patent only uses load data and does not consider photovoltaic power generation, wind power generation, and electric vehicles. Charging load, and even real-time environmental data such as local temperature, light, wind speed, etc., the data source has a single form, which cannot effectively represent the actual operating status of the distribution network, and fails to make full use of the information contained in the massive database, which is inconsistent with the actual project

At the same time, in this patent, the random matrix is ​​constructed with a time window of 24 o'clock in a single day, the sampling time interval is 1 hour, and the number of sampling times is 24. The random matrix constructed by this method cannot meet the needs of real-time reactive power optimization of the distribution network, and is different from the actual Engineering does not match

Moreover, in this patent, only the average spectral radius is extracted to approximate the distribution characteristics of the matrix data. The form of the statistical feature quantity is single, and the statistical distribution law of the eigenvalues ​​cannot be fully extracted, which will lead to poor reactive power optimization control sequence and affect the Reactive power optimization effect of distribution network

Finally, this patent does not propose a specific and effective matching method to achieve distribution network scenario matching

Also published by Liu Keke et al., the patent "A Method for Determination of Reactive Power Optimal Control Sequence of Distribution Network Based on Big Data", based on the previous patent, proposes to calculate the correlation between historical load and current load based on the average spectral radius coefficient, by comparing the size of the correlation coefficient, obtain the load day with greater correlation, and take the reactive power optimization control sequence of that day as the current optimization sequence. The demand for power optimization, the effect of reactive power optimization is poor, and the selected data source and extracted statistical feature quantity are single, which cannot effectively represent the operation status of the actual distribution network, which is inconsistent with the actual project

Images