接触网复合绝缘子污闪电压预测方法研究

Research on prediction method of flashover voltage of composite insulators of contact network

通过人工污秽实验室中对接触网常用的11种复合绝缘子进行大量污秽实验,提出一种基于改进生物地理学算法优化BP神经网络的接触网复合绝缘子临界污闪电压预测模型。选取伞径Dm,爬电距离L,形状因数f及等值附盐密度ESDD作为模型的输入特征量。试验和对比分析表明,模型的预测效果良好,预测结果与试验结果基本一致。预测值可作为污闪试验的参考数据,能够有效降低污闪试验的工作量,为评估接触网线路的可靠性及对新建线路外绝缘的选型和维护提供一条新思路。

11 kinds of composite insulators commonly used by the contact network undergo a series of pollution flashover experiments in the artificial fog cabinet.Following that,a critical pollution flashover voltage prediction model of composite insulators of the contact network was proposed based on optimization of BP neural networks and an improved biogeographic algorithm was put forward.Cap diameter Dm,creep distance L,form factor f,and equivalent salt deposit density,ESDD,were adopted as input characteristic quantities of the model.Experimental and comparative analysis results suggest the model’s prediction effect is favorable,and that prediction results basically coincide with experimental results.The prediction value can be adopted as the referential data of the flashover experiment and can efficiently reduce the workload of the pollution flashover experiment.This research can provide a new perspective for reliability evaluation of contact network lines and model selection and maintenance of external insulation of newly-built lines.