基于ReliefF-NOSCA-AdakNN的肌肉疲劳识别技术研究

针对竞技体育训练中的肌肉疲劳监测问题,提出了一种基于ReliefF-NOSCA-AdakNN（RNA）的表面肌电信号（sEMG）特征提取和分类算法. 该算法结合了特征和类别之间的相关性分析和启发式搜索算法,对高维特征进行了有效的筛选和分类. 将RNA算法应用于经过滤波处理的肱二头肌肌电信号数据,对不同疲劳状态进行了识别和分类. 实验结果表明,提出的RNA算法在平均分类准确率和标准差方面分别达到了83.88%和0.012 7,均显著优于传统单一算法,体现了较好的分类性能.

Research on muscle fatigue identification technology based on ReliefF-NOSCA-AdakNN

To address the problem of muscle fatigue monitoring in competitive sport training, a surface electromyography （sEMG） feature extraction and classification algorithm based on ReliefF-NOSCA-AdakNN （RNA） was proposed. The analysis of the relevance between features and classes with the heuristic search algorithm, were combined by this algorithm. Besides, high-dimensional features were effectively selected and classified. The RNA algorithm was applied to the filtered biceps brachii sEMG data to identify and classify different fatigue states. The experimental results showed that the proposed RNA algorithm significantly outperformed the traditional single algorithms in terms of average classification accuracy and standard deviation, which reached 83.88% and 0.0127 respectively, demonstrating a good classification performance.