论文部分内容阅读
为了提取泄漏电流安全区内体现绝缘子污秽度的特征信息,在反复试验的基础上,测取了4种污秽度下XP-70型绝缘子串的泄漏电流,采用了自适应阈值法对原始信号进行消噪,提取出安全区内的3个特征量:有效值均值、有效值最大值和有效值标准差,分析得出这3个特征量与绝缘子表面等值附盐密度呈指数拟合关系。选择了400组泄漏电流有效值特征量作为训练与识别样本,建立了基于概率神经网络的绝缘子污秽度预测模型。仿真和试验结果对比表明,模型预测准确率达87.5%。研究结果不仅对线路绝缘子清扫决策有一定的指导作用,而且为输电线路污闪预警系统特征量的优化选择提供了理论依据。
In order to extract the characteristic information of insulator contamination in the safety zone of leakage current, leakage current of XP-70 insulator string under four pollution levels was measured on the basis of trial and error. The adaptive threshold method was adopted to process the original signal Denoising and extracting three characteristic quantities in the safe area: the mean of effective value, the maximum value of effective value and the standard deviation of effective value. The analysis results show that the three characteristic quantities are exponentially fitted to the salt density of insulator surface equivalent. 400 groups of leakage current rms features were selected as training and identification samples to establish a predictive model of insulator contamination based on probabilistic neural network. The simulation and test results show that the model prediction accuracy rate of 87.5%. The results not only provide some guidance to the decision of the line insulator cleaning, but also provide the theoretical basis for the optimization selection of the characteristic quantities of the flashover warning system of the transmission line.