摘要
针对传统电力信号识别算法中特征选取的随意性,提出了一种基于遗传支持向量机(GA-SVM)的电压暂降信号识别方法。首先通过S变换时频分析法提取该信号识别需要的可能特征集,然后利用遗传算法的全局搜索特性得到优秀特征,最后通过多分类支持向量机实现暂降信号识别并验证选取特征的有效性。仿真结果证明,该方法能快速、有效识别出电压暂降信号类型。
In order to reduce the randomness of feature selection in the traditional power signal identification algorithm,a new approach for identification of voltage sag signal is proposed based on genetic algorithm optimized support vector machine（GA-SVM）.First,the initial feature set is extracted with S-transform time-frequency method,which is probably used for identification of voltage sag signal.Then the GA is used for search of the key features from the feature set.Finally,the multi-classification SVM is applied for identification of voltage sags signal and prove of the validity of the selected features.The simulation results show that the approach could identify the voltage sags signal effectively and quickly.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2012年第1期84-87,共4页
Proceedings of the CSU-EPSA
基金
国家自然科学基金资助项目(51107120)
浙江省自然科学基金资助项目(Y1090182)
浙江省教育厅科研项目(Y201120550)
关键词
电压暂降信号
遗传算法
支持向量机
特征选取
识别
voltage sag signal
genetic algorithm
support vector machine
feature selection
identification
