【摘要】 外骨骼机器人发展迅速，基于生理信号的运动意图识别在人机协同控制研究中得以重视。针对肌电信号易受肌肉疲劳影响和采集要求高的缺点，提出一种基于肌音信号的核主成分分析和改进支持向量机(KPCAGASVM)的模式识别方案，对平地行走、上楼下楼和上坡下坡5种步态进行模式识别研究。基于遗传算法进行参数调优，其识别方案KPCAGASVM的识别准确率为97.33%,优于PCAGASVM和其他分类器。实验验证，基于肌音信号的KPCAGASVM为一种高效的步态运动识别方案。更多还原

【Abstract】 Exoskeleton robots are developing rapidly, and the recognition of motor intent based on biological signals has been valued in human-machine collaborative control. To solve the disadvantages that EMG signal is susceptible to muscle fatigue and the acquisition requirements are high, a pattern recognition scheme based on Kernel Principal Component Analysis(KPCA) and improved Support Vector Machine(SVM) based on mechanomyography(MMG) signal is proposed to study five kinds of gait pattern recognition, including flat walking, going upstairs, going downstairs, going uphill and downhill. Based on the parameters tune of the GA algorithm, the recognition accuracy of KPCAGASVM is 97.33%, which is better than that of PCAGASVM and other classifiers Experiment results show that KPCA-SVM based on MMG signals is an effective gait recognition scheme.更多还原