Research On Low-Frequency Oscillation Safety Early Warning And Situational Awareness Prediction Mechanism For New Power System

The low-frequency oscillation problem in the new power system is one of the important factors affecting the safety and stability of China’s power grid.Conducting rapid and accurate warning research on low-frequency oscillation phenomena,and providing reliable and real warning indicators for dispatching personnel,is of great significance for the safe and stable operation of the new power system.Firstly,aiming at the problem of low accuracy in current low-frequency oscillation warning strategy identification,a method for identifying low-frequency oscillation amplitude warning indicators based on key feature wide-area dimensionality reduction data Vinnicombe distance is proposed for new power system.This method first pre-processes and screens the original big data collected by PMUs,extracts key feature data,and generates the initial feature quantity matrix required for low-frequency oscillation amplitude warning indicators.Then,by numbering the monitoring terminal nodes of the new power system and the regions between each node,the corresponding association relationship and correlation value are determined,generating a network association multi-feature vector state detection matrix.Combining with the Vinnicombe distance calculation to determine the transfer function distance,it judges whether low-frequency oscillation occurs and effectively improves the identification accuracy of low-frequency oscillation amplitude early warning.Finally,an example simulation is conducted on the New England 10-machine 39-node system and compared with the traditional LOF fault location method.The proposed method has a higher accuracy compared with the traditional method,with a 14.82% reduction in error.Secondly,aiming at the problems of single warning indicator and poor accuracy of traditional low-frequency oscillation warning strategy in new power system,and the lack of online warning system,this chapter proposes a method to select and extract key data from the raw data collected by PMU to form multiple warning indicators.A trapezoidal fuzzy number judgment matrix is generated,and the membership functions of each warning indicator are constructed after verifying the kernel matrix.A comprehensive warning indicator is calculated and classified using a random forest algorithm for the causes of low-frequency oscillation in new power system,thus building a low-frequency oscillation system.Simulation verification is conducted using the New England 10-machine 39-node system.Compared with the traditional Analytic Hierarchy Process(AHP)method,the proposed method can effectively reduce the impact of strong subjectivity caused by an expert judgment matrix,and solve the low-frequency oscillation warning problem in the new power system with multiple indicators.Next,to address the issue of situational awareness in new power system,an improved particle swarm optimization algorithm is utilized to optimize the random forest model,which is used to predict the processed data of various components of the PMU.An analysis is conducted on the accuracy of prediction results and the computational time required by the model through numerical simulations.In this study,the prediction time of our method is 7.03 seconds,which is much lower than the other three methods.The prediction accuracy is95.43%,which is higher than the other three methods.Finally,this paper develops a visual warning platform software with real-time data detection and timely alarm function combined with practical applications.Users can have their own accounts and discover and solve problems in a timely manner,reducing a lot of time and economic losses.The platform allows remote control of various equipment parameters and data,and has excellent human-computer interaction.The simulation results of this study show that in the event of low-frequency oscillations occurring in power system with new energy sources,various warning indicators can quickly and accurately provide information about the safe operation status of the power system.This verifies the feasibility and correctness of the safety warning indicator analysis mentioned above and provides theoretical and technical support for low-frequency oscillation warning issues in China’s new power system.