基于ReliefF算法与遗传算法的肌电信号特征选择

针对肌电信号特征维数高、运算效率低等问题,提出了一种基于ReliefF算法与遗传算法(GA)相结合的肌电信号特征选择方法.分析了肌电信号的特征,运用小波分析对肌电信号进行特征提取,采用ReliefF算法评估所提取的高维特征信号的权值,以选出对分类效果影响显著(权值较大)的特征子集,采用GA进一步筛选出分类效果最佳的特征子集,并对比分析了基于ReliefFGA-Wrapper算法与全局搜索算法对肌电信号处理的时间和分类效果.结果表明,所提出的方法能够提高运算效率并具有很好的分类效果.

Feature Selection of Emg Signal Based on ReliefF Algorithm and Genetic Algorithm

Abstract： To address the high dimension of signal characteristics and low operation efficiency of electromyography (EMG), an algorithm for feather selection was proposed based on ReliefF feature evaluation and the genetic algorithm. The characteristics of the signal was analyzed, the features of the EMG signal with wavelet transform were extracted, the weight of each feature was assessed using the ReliefF algorithm, and the feature subset which has a obvious influence upon classification result was selected. Then the best feature subset for the classification results was screened out by using the genetic algorithm. Besides, the operation time and classification results of the ReliefF GA Wrapper algorithm were compared with those of the global search. The result shows that the proposed algorithm not only guarantees a good classification result but also improves the operational efficiency.