【摘要】 通过腐蚀试验获得60组110 kV变电站土壤成分数据及相应的金属片腐蚀速率,对60组样本数据进行灰色关联分析,结果表明Cl~-含量、pH值、含水量和含盐量是接地网腐蚀的主要原因,并以此作为接地网腐蚀预测模型选择支持向量的依据。为提高最小二乘支持向量机模型的预测精度,采用扩展记忆粒子群算法对最小二乘支持向量机的惩罚因子和核函数参数进行寻优,建立基于PSOEM优化LSSVM的接地网腐蚀预测模型。应用试验数据进行仿真分析,结果表明,PSOEM-LSSVM模型在训练拟合和外推预测方面效果更好。更多还原

【Abstract】 The data of soil composition and the corrosion rate of metal sheet in 60 groups of 110 kV substation is obtained through corrosion test. These sample data are analyzed by grey relation,showing that Cl-content,pH value,water content and salt content is the main reasons for the corrosion of grounding grid,based on which support vectors are selected for a prediction model. In order to improve the prediction accuracy of LSSVM model,particle swarm optimization with extended memory is used to optimize the penalty factor and kernel function parameters of the LSSVM. The corrosion prediction model of the grounding grid based on PSOEM-optimized LSSVM is established,and the simulation is done with the test data. The results show that the PSOEM-LSSVM model is better in training and fitting and prediction by extrapolation.更多还原