摘要
针对电能质量扰动识别问题,提出一种多级相关向量机(RVM)和提升小波包分解(LWP)相结合的扰动分类新方法。根据电能扰动现象的内在特征,首先通过提升小波包算法快速提取各类扰动信号的分解系数能量作为扰动特征量;然后利用相关向量机构建多级分类树模型实现分类识别任务。研究表明相关向量机在权系数上引入超参数,与支持向量机相比无需设置惩罚系数、推广能力好、解更稀疏。仿真表明所采用方法能够快速有效地获取高精度扰动分类识别率,测试时间短,更适合于在线检测。仿真和试验结果验证了所采用方法对电能质量扰动分类的有效性。
A new method classifying power quality disturbances based on lifting wavelet packet decomposition and multi-lay relevance vector machine（RVM） is presented.According to the intrinsic characteristics of power quality disturbances,samples are decomposed by lifting wavelet packet（LWP） algorithm using second-generation fast lifting wavelet transform,and the energy of the LWP coefficients of each end node are extracted as eigenvectors.Then the disturbance types are identified through the multi-lay RVM pattern recognition classifier.As a sparse Bayesian learning algorithm,RVM doesn＇t need penalty factor parameter,constrains the weight coefficient using hyper parameter,and leads to sparser model with better generalization ability compared with support vector machine（SVM）.Simulation results show that the proposed LWP-based RVM method can achieve higher classification accuracy quickly,requires substantially fewer relevance vectors and shorter test time than the SVM classifier.The numerical result verifies its validity to classify power quality disturbances.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2010年第3期782-788,共7页
High Voltage Engineering
基金
江苏省科技攻关项目(BE2007069)
镇江市社会发展项目(SH2008005)
江苏大学高级专业人才科研启动基金(07JDG072)~~
关键词
电能质量
扰动分类
相关向量机
支持向量机
小波包分解
提升算法
power quality
disturbance classification
relevance vector machine（RVM）
support vector machine（SVM）
wavelet packet decomposition
lifting algorithm
