Power System Fault Classification And Prediction Based On The Artificial Intelligence And The Data Mining

In recent years,as the structure of the power system changes,its safe operation is related to the development of the entire national economy.In order to ensure the reliability and stability of the power system,predicting upcoming power faults in advance and taking corresponding preventive measures can effectively prevent the occurrence of the abnormal electricity accidents and reduce the economic losses.However,due to the lack of the precision and the accuracy of the traditional power grid protection methods,the protection device may malfunction or refuse to operate,therefore,the study of the new power system fault diagnosis methods has attracted widespread attention from the domestic and foreign scholars.Nowadays,with the rapid development of the information technology and the rapid advancement of the smart grids,the fault classification and prediction method based on the artificial intelligence and data mining has become a brand new field of the power system protection.The innovation and main research work of this article includes: Firstly,a distribution network fault classification and prediction method based on the clustering,association rules and stochastic gradient descent regression is proposed.First of all,Kmeans clustering is used to preprocess the source data derived from the IEEE 9 fault model built in the electromagnetic transient simulation software PSCAD;Then the association rules filter the samples in the sample library to dig out the samples whose correlation is higher than the set evaluation index in advance,these samples are regarded as the associated sample library;Finally,the associated sample library is used for regression to obtain the classification and prediction model,and the experimental result shows that the model can accurately and efficiently classify and predict the distribution network faults.Secondly,a transmission line fault prediction model based on BP neural network optimized by adaptive moment estimation is studied,and the weight value and bias value of the neural network are adjusted and optimized.Based on the transmission line fault data provided by a power company,the stochastic gradient descent optimization model under the same conditions is used for comparison,so a combined algorithm use strategy of the two models is formulated.Then it is verified through the experiment that the prediction accuracy of the combined algorithm model is higher than that when the algorithm is used alone,which proves the feasibility of the proposed strategy.