基于声发射波关联维数的绝缘子放电识别研究  

Identification Research of Insulators Discharge Based on Correlation Dimension of Acoustic Emission Wave

作　　者：李光 王成江[2] 李如锋[2] 范开明[2] 胡帅[2] 张倩 

机构地区：[1]山东临沂供电公司,临沂276000 [2]三峡大学电气与新能源学院,宜昌443002 [3]滕州富源低热值燃料热电有限公司,滕州277523

出　　处：《高压电器》2013年第5期79-85,共7页High Voltage Apparatus

摘　　要：以绝缘子放电声发射波的时域主成分值作为特征量,进行常见3类放电的识别研究时,因放电间歇性、放电声波中数据坏点的影响,导致放电识别效果欠佳。故笔者将绝缘子放电声发射波看作一分形体,提取其分形维数作为识别特征量。具体选取了在绝缘子3类放电上分布差异较大的分形关联维数作为识别放电的特征量,通过研究测试,总结了各类放电关联维数的分布范围及规律。最后,设计BP网络,进行识别测试,结果表明,与时域主成分值相比,以关联位数作为识别特征量,更能准确、有效地识别绝缘子的放电类型,识别正确率达到89%左右。When conducting the common three types of insulators discharge recognition research with principal component values of the time-domain discharge acoustic emission wave as characteristics, clue to the impact of discharge intermittent and dead point in the acoustic emission wave, the discharge recognition effects perform badly. Therefore, discharge acoustic emission wave is seen as a fractal and its fractal dimensions are extracted in this paper. Speeially,fractal correlation dimension which shows great difference on the three type insulator discharge is selected as characteristic to identify discharge. Through some research and testing, the distribution scope and regularity of correlation dimension for various types of discharge are summed up. Finally, identification test is carried out with BP network, and from the results, compared with the time-domain principal component values, as identifying characteristics, correlation dimension could identify the type of insulator discharge more accurately and effectively, and the identification rate has reached 89% or so.

关 键 词：声发射 特征量 分形 关联维数 放电识别 

分 类 号：TM216[一般工业技术—材料科学与工程]