Study On Generating Stable Operation Rules Of Power System Based On Association Analysis

The safe and stable operation of the power system is related to the national economy and the people's livelihood.The online safety assessment of the power system has always been the focus of research by experts and scholars.With the rapid development of China's power grid,the scale of power grids continues to expand,the uncertainty of power flow distribution is gradually increasing.and the stability problems are complex and diverse.Higher requirements are placed on the online safety assessment and real-time control capabilities of the power grid.The traditional safe production methods based on manual experience and online simulation calculations arc difficult to meet the higher requirements for scheduling control capabilities.The traditional safe production methods based on manual experience and online simulation calculations are difficult to meet the higher requirements for scheduling control capabilities.Relying on the development of communication,computer technology,and accumulated power big data,this paper applies machine learning technology to the discovery and extraction of scheduling operation stability rules.Through the probability statistics of historical data events,the intrinsic relationship between the operation mode and stability is found.This relationship may be well known to the operator or may be an undiscovered potential law.Firstly,this paper sorts into a stable evaluation database by extracting the steady-state feature quantity of the power system and the safety and stability level obtained by simulation analysis.Comparison of integrated learning algorithms:the application of random forest algorithm and XGBoost algorithm in feature extraction process,according to the characteristics of data sample set imbalance,select XGBoost algorithm to extract key features.Using PSD-BPA simulation software,batch simulation analysis of IEEE39 node system is carried out to generate a large sample set,and the effectiveness of the proposed feature selection method is verified.Then,since the association rule model is based on discrete data,the steady-state feature quantity of the system collected in this paper is mostly continuous data.Therefore,the feature discretization algorithm of weighted k-means clustering combined with information gain is proposed,and the influence of different features on clustering results and the correlation between feature attributes are considered.The discretization results are evaluated according to the evaluation index,and the method is verified by IEEE39 node system.It is proved that the proposed method can obtain reasonable discrete intervals.Finally,the association rule analysis method based on FP-Growth algorithm is applied to generate stable operation rules of power system.Based on the processing of the sample data in the previous chapters,a stable evaluation rule base is constructed.While using the IEEE39 node system simulation verification,the historical data of the actual large-grid online safety assessment system is collated,and static stability constraints and dynamic security constraints are considered to generate stable operation rules.