摘要
针对高压断路器操作机构状态评估过程中线圈电流特征量维数过高的问题,提出了一种基于皮尔森相关系数的特征量优化方法。利用提取的线圈电流特征量构造皮尔森相关系数矩阵,通过分析线圈电流8个特征量间的相关性,并进行特征量优化,得到具有较高分类能力的3个特征量。将优化后的特征量输入神经网络进行高压断路器操作机构的状态评估。实例分析表明特征量优化方法有效降低了特征量的维数,简化了分类器结构,以较少特征量达到了较理想的高压断路器操作机构状态评估效果。
Based on the Pearson correlation coefficients, a new method for optimizing characteristic quantity is proposed to reduce the characteristic dimension of coil current in status assessment process of high voltage circuit breaker operating mechanism. To obtain three characteristic quantities with higher classification capability, the extracted coil current characteristic quantity is made use of to construct the Pearson correlation coefficient matrix, and the correlations of eight coil current characteristic quantities are analyzed and optimized. Then, the optimized characteristic quantities are input to neural network to assesse the status of high voltage circuit breaker operating mechanism. Experimental results show that the proposed method reduces the dimension of characteristic quantities, simplifies the structure of classifier, and achieves satisfactory status assessment of high voltage circuit breaker operating mechanism with fewer characteristic quantities.
出处
《高压电器》
CAS
CSCD
北大核心
2016年第6期199-203,共5页
High Voltage Apparatus
关键词
高压断路器
线圈电流
特征量优化
状态评估
high voltage circuit breaker
coil current
characteristic quantity optimization
status assessment
