Bootstrap方法在局部放电特征提取中的应用

The Application of the Partial Discharge Feature Extraction Using a Bootstrap Approach

介绍了用Bootstrap方法对局部放电特征的小样本原始数据集重复再采样 ,提取其特征参量的统计信息 ,建立绝缘缺陷特征隶属函数 ,并应用遗传编程分类方法对缺陷模型进行模式识别的方法。试验结果表明该方法有效、可行。

Fuzzy attributes of the insulation defects are used in the study of the automatic PD pattern recognition for insulation defect models. In order to eliminate the bottleneck in knowledge acquaintance based on a small sample data, this paper presents the use of a non-parametric bootstrap method to extract PD features of insulation defects. The statistical information of insulation defect parameters is characterized by this approach, and the membership functions for PD data set are constructed. Moreover, a new method of PD pattern recognition, genetic programming algorithm, is proposed. The experimental results indicate that this method is effective and feasible.