基于改进支持向量机的绝缘子污秽度在线监测

Insulator pollution degree online monitoring based on improved support vector machine

针对传统绝缘子污秽度监测方法的不足,基于石英棒的光损耗原理,设计了光纤污秽传感器,通过实验测得温度、湿度、光功率衰减百分比与盐密的关系。利用改进的支持向量机对样本进行训练,通过大量数据比对,验证了该改进算法的准确性,较传统模型对数据的拟合度更高。在此基础上研制了污秽度监测终端,能进行温湿度及光功率数据的采集,并将采集的数据发送至后台。开发了输电线路绝缘子污秽度在线监测系统,可实现对采集数据的计算分析,得到实时盐密值及污秽度并绘制曲线图。测试结果表明,该系统实现了绝缘子污秽度的在线监测,为运行人员提供了参考。

Aiming at the shortcomings of traditional insulator pollution monitoring method, based on the optical loss of quartz rods, designed fiber optic contamination sensor, measured the relationship between temperature, humidity, percentage of luminous flux decay and equivalent salt deposit density （ESDD）. Using the improved support vector machine （SVM） to training model, through a large number of data comparison, the accuracy of the improved algorithm is verified, compared with the traditional model, the fitting degree of data is higher. On this basis, developed a pollution degree online monitoring terminal, it can collection the temperature and humidity and luminous flux data, and send the collected data to the background. Developed the on-line monitoring system of transmission line insulator pollution degree on-line monitoring system, to achieve the calculation and analysis of the collected data, get the real- time ESDI） and draw the curve. Test results show that, this system achieves online monitoring of insulator pollution degree, provide a reference for the operators.