一种接触网绝缘子污闪预测方法的研究

摘要

Frequent occurrence of insulator pollution flashover of contact line in recent years severely threatens the safety of railway transportation.In order to provide scientific and efficient guidance for anti-pollution flash-over,this paper proposed the prediction of the pollution grade and pollution flashover probability of insulators in contact line.Firstly,the BP neural network was optimized by using artificial fish swarm algorithm to estab-lish the forecasting model of insulator pollution grade.Four representative meteorological factors such as monthly precipitation,monthly average wind speed,monthly average relative humidity as well as monthly dust-fall volume,and insulator running time(month)were selected as input characteristic quantities.The data of five characteristic quantities in last month were input to forecast the insulator pollution grade for this month. Then,the forecasting model of pollution flashover probability based on insulator pollution grade was estab-lished to predict current month's pollution flashover probability of insulators in the contact line.Early warning can be conducted and a short-term forecast response mechanism for circuit pollution flashover in contact line was established.It offered a new thought for the evaluation of the reliability of contact line circuit,the develop-ment of effective anti-pollution flashover strategy by operation and maintenance departments and the selection of insulator type as well as maintenance of newly built circuit insulators.%近年来接触网绝缘子污闪事故频发,严重威胁铁路运输安全生产.为了给防污闪工作提供科学高效的指导依据,本文提出对接触网绝缘子污秽等级和污闪概率进行预测.利用人工鱼群算法优化 BP 神经网络,建立绝缘子污秽等级预测模型.选取月降水量、月均风速、月均相对湿度、月降尘量等4个最具代表性的气象因子和绝缘子运行时间(月)作为输入特征量.输入上个月5个特征量的数据,预测当月绝缘子的污秽等级.建立基于绝缘子污秽等级的污闪概率预测模型,预测当月接触网绝缘子污闪概率,提前进行预警并建立接触网线路污闪短期预报响应机制.为评估接触网线路的可靠性和运维部门制定有效的防污闪策略以及对新建线路绝缘子的选型和维护提供一条新的思路.