摘要
现有的特高频局部放电定位法主要是基于时延进行定位,这类方法的硬件价格高,实现较为困难。该文提出基于压缩感知技术和接受信号强度(RSSI)指纹图谱的局部放电定位法,具有易实现和较强的环境适应性等特点。实际定位时先构建测试现场的RSSI指纹图;再使用BP神经网络对局部放电源进行初定位,并以此为基础构建缩小的RSSI指纹图;最后通过压缩感知技术进行精确定位。该方法利用神经网络和压缩感知两种算法的优势,兼顾了定位效率和精度的要求。现场测试结果表明,基于压缩感知的局部放电定位法平均定位误差为0.2 m,且93.9%的定位误差≤1 m,验证了所提出的定位方法的精确性,该方法具有较好的实际应用价值。
Existing UHF partial discharge (PD) localization technology is mainly based on time delay algorithm which has expensive hardware cost and is hard to achieve. This paper proposed a compressed sensing based PD localization methodology which was based on received signal strength indicator (RSSI) fingerprinting localization algorithm. It has easy fulfilling features and good environmental adaptability. Firstly, the RSSI fingerprinting was built in the offline stage. Secondly, in the online stage, BP neural networks is used to achieve preliminary localization, then compressed sensing strategy is deployed to achieve more accurate localization. The filed test showed that the mean errors of PD localization by out proposed method is 0.2 m and 93.9% of errors is smaller than 1 meter. The test proved that proposed localization algorithm is accurate and has good practical value.
出处
《电工技术学报》
EI
CSCD
北大核心
2018年第1期202-208,共7页
Transactions of China Electrotechnical Society
基金
国家电网公司科技项目资助
关键词
局部放电
定位
压缩感知
接受信号强度
指纹图
神经网络
Partial discharge, localization, compressed sensing, received signal strength indicator, fingerprint, neural networks