Research On Corrosion Rate Prediction Method Of Grounding Grid

In the past few years,with the rapid development of power system,the country's power grid scale and system capacity has the obvious promotion.At the same time,the use of various kinds of secondary regulation and control equipment in the system is more and more.If the grounding grid has been bad,it's easy to cause serious accidents.Prediction of the corrosion rate of grounding grid can accurately grasp the state of corrosion of grounding grid without excavation,which reduces the workload of corrosion inspection of grounding grid and provides an important reference value for anti-corrosion measures.So the research of corrosion rate prediction method for grounding grid has important practical significance.In the process of the building for prediction model sample library,made the corrosion diagnosis of grounding grid in the first place using the diagnosis method for corrosion of grounding grid based on the electrical network theory.According to the result of corrosion diagnosis,put forward that choose the certain branch area after corrosion diagnosis as a sampling point,which improves the rationality of selecting the corrosion sample points.At the same time,this paper put forward the growth rate of conductor resistance as a new feature,which was together with soil resistivity,moisture content,porosity,the content of SO₄^2-and Cl^-of the soil in sampling points as the influence factors of corrosion rate.After the collection of these factors parameters,a sample library for corrosion prediction of grounding grid was set up.Aiming at the limitations of the artificial bee colony algorithm,the algorithms of the various stages of the strategy were improved,and the YHABC algorithm was proposed.According to the results of performance testing,the optimization precision and convergence speed of the YHABC algorithm were better compared with the basic ABC algorithm,PSO algorithm and DE algorithm.A corrosion rate prediction method for grounding grid was established based on the SVM theory.At the same time in order to improve the accuracy of prediction method,the SVM model penalty factor C and the kernel function parameter?were optimized using YHABC algorithm,and put forward the corrosion rate prediction method base YHABC-SVM.Test results showed that the prediction model based on YHABC-SVM possess good accuracy and stability,which had better performance than prediction model based on BP neural network.This paper introduced the combination prediction method.Firstly,a single prediction method using generalized regression neural network was established.Then another single prediction method using extreme learning machine optimized by artificial fish swarm algorithm was established.Then according to the basic theory of the IOWHA operator,this paper put forward that take the average prediction accuracy of the samples near the prediction sample point as the induced value of single prediction model.On the basis of two kinds of single prediction method,the combination prediction method of corrosion rate for grounding grid was established based on IOWHA operator.Finally,the testing index indicated that the error and volatility of the combination prediction method were lower compared with two kinds of single forecasting method,which embodied the combination prediction method for grounding grid corrosion rate based on the IOWHA operator possess excellent properties.