基于正态模型的电气试验方法优化及参数估计研究

为掌握正态分布电气特性数据的正确试验方法,确定合理的参数估计值,文章基于最小二乘参数估计方法,针对多级法和升降法电气试验数据进行中值及标准偏差统计分析。研究结果表明,多级法与升降法试验数据的中值估计值50%放电电压U50有很好的一致性,仅相差3. 7%;而两种试验方法对应的标准偏差估计值相差较大,达到42. 6%,因此,在实际电气试验工作中,尤其对于获取大尺度绝缘的标偏试验,应优先选择多级法,而获取50%放点电压则两种方法均可以满足要求。

Research on Electrical Test Method Optimization and Estimating Parameters Based on Normal Model

In order to correctly select the test method which conforms to the normal distribution model and determine the reasonable statistical value of parameters, based on the least squares parameter estimation method, this paper makes a statistical analysis of the median and standard deviation of the test data of multi-stage method and lifting method commonly used in electrical tests. The results show that the two test methods have good consistency with the corresponding median U50 estimation value, but the difference between two statistical standard deviations is large, reaching 42.6%. Therefore, in the test work, especially for the parameter statistics of large-scale insulation test for obtaining standard deviation, multi-stage method should be preferred, not lifting method. Both methods all can meet the requirements of obtaining 50% discharge voltage.