基于自组织神经网络的紫外电晕检测状态评定

针对传统紫外成像法的局限性,提出了一种有效进行紫外电晕检测的状态评定方法.通过设计构建的单通道紫外成像系统采集的图像信息,结合环境因素,提取了在一定时间间隔两种不同电晕放电状态的电子崩性放电面积、流注性放电面积、流注性放电重复次数以及湿度的4个特征参数,并在此基础上,建立了自组织特征映射神经网络的高压设备电晕状态评估模型.测试结果达到预期效果,具有良好的准确性和稳定性,为高压设备的运行状况诊断带来实际意义.

Ultraviolet Corna Detection Status Evaluation Based on Self-Organizing Neural Network

An effective evaluation method of ultraviolet corona detection that beyond the limitations of traditional ultraviolet imaging method is put forward.With the image information acquired by the built single channel ultraviolet imaging system,combined with the environmental humidity,the electron avalanche discharge area,lingers discharge area and the number of repeated discharge are taken as the four characteristic parameters of the inputs.And on the basis of the advantages,the self-organizing neural network model of high voltage equipment corona state evaluation with good accuracy and stability is established by using SOM clustering evaluation as tested on Matlab platform.The good results bring practical significance for high voltage equipment running status diagnosis.