论文部分内容阅读
气体绝缘电器(GIS)局部放电(PD)信息的传统模式识别方法存在识别率低、特征维数过高、信息利用不充分等问题。为此,根据Dempster-Shafer证据理论,将对局部放电的时间解析信号分析模式和局部放电的相位解析信号分析模式的融合决策诊断引入GIS故障分析,实现了对GIS内部绝缘缺陷的有效综合判别。仿真与实验结果表明,该方法能深入挖掘识别中的有效信息,并通过一定的决策规则实现对信息的总体描述,具有良好的拒误诊断能力。此外,基于2种分析模式对4种典型缺陷的综合识别率均高于80%,且互补性较强,在降低特征维数的同时显著提高了识别准确度。
The traditional pattern recognition methods of partial discharge (PD) in gas insulated electrical appliances (GIS) have some problems such as low recognition rate, high characteristic dimension and insufficient information utilization. Therefore, according to the Dempster-Shafer evidence theory, the fusion fault diagnosis of time-resolved signal analysis mode of partial discharge and phase-resolved signal analysis mode of partial discharge is introduced into GIS fault analysis, and the effective comprehensive discrimination of insulation defects in GIS is achieved. The simulation and experimental results show that this method can dig out the effective information in recognition deeply, and realize the overall description of the information through certain decision rules, which has a good ability to diagnose and reject errors. In addition, the comprehensive recognition rate of the four kinds of typical defects based on the two analysis modes is higher than 80% and is highly complementary, which remarkably improves the recognition accuracy while reducing the feature dimension.