Study On Salt Density Prediction Method For Insulator Of Transmission Line

With the rapid development of China's economy,the scale of power grid is increasing,and people's demand for the reliability and safety of electricity is increasing.However,the intensification of human activities is accompanied by the aggravation of environmental pollution.The surface pollution of transmission line insulators is increasingly serious,and the frequent occurrence of power accidents causes great losses to the people's economy.Conductive material on the surface of a transmission line insulation study found that salt density is the main material of transmission lines occur leakage,but also an important measure of transmission line insulator pollution.With the rapid development of smart power grid,more salt density analysis technology has emerged,among which the salt density prediction has become the main research field as a direct and effective way.Salt density prediction method are mostly point prediction at present,the main use of support vector machine,artificial neural network,such as the mainstream,and salt density range prediction and probability density prediction and other fields also few people involved.In this paper,the characteristics of transmission line insulator salt density data analysis,modeling of salt density prediction based on the salt density data,especially for salt of interval prediction and probability density prediction problems were studied,the main research content is as follows:(1)Analyzed the salt density data is chaotic,completed the salt close time series phase space reconstruction,and introduces quantile thought and realization method,for the prediction of transmission line insulator salt density research provides theory basis.(2)The mathematical model of traditional prediction method is introduced,the advantages of the neural network quantile regression model are analyzed,and the salt density prediction model of RBF neural network quantile regression is established.(3)Discovered the limitations of traditional AIC criterion of salt density data,and improve the AIC criterion is used to analyse the predictive model parameters choice,at the same time by using particle swarm optimization(pso)algorithm and genetic algorithm to optimize the model parameters,and the results of simulation analysis,choose the most suitable optimization algorithm,prove the validity of the improved,obtained the optimum salt density prediction model.(4)The salt density probability density estimation model was established,and the kernel density estimation model was established based on the salt density value of the prediction of neural network quantile regression.Introduces the commonly used kernel function and window wide selection method,try to six kinds of combination method to build the kernel density estimation model,and choose the optimal combination of methods as the optimal kernel density estimation model;The prediction error of the method used in this paper is compared with that of traditional salt density prediction method,and the simulation analysis is carried out to prove that the method chosen in this paper is more effective.