融合增强学习机制的智能假肢肌电控制方法

项目名称： 融合增强学习机制的智能假肢肌电控制方法

项目编号： No.51205080

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 机械工程学科

项目作者： 杨大鹏

作者单位： 哈尔滨工业大学

项目金额： 24万元

中文摘要： 本项目以实现长期有效的智能假肢多运动模式肌电控制为目标，通过开展肌电信号时变特性表征方法、学习型肌电识别系统建模方法及相关计算学习理论、具体算法以及实验研究，重点突破时变肌电信号中运动信息的特征表现形式、增强知识学习与肌电模式识别的信息融合机制、增强型肌电模式学习的效率及收敛性等关键科学问题，为新型智能生机电假肢控制系统提供设计新原理与新方法。本项目的进行将有助于加速我国灵巧假肢产业化进程，提升康复医疗装备技术水平，并对新型智能生机电一体化系统的设计具有一定借鉴意义。

中文关键词： 灵巧假手；肌电控制；机器学习；模式识别；

英文摘要： For achieving a long-term efficient EMG control method for the multi-DOFs prosthetic hand, this project carries out researches on the character representation of the time-variant EMG signals, modeling and fundamental calculating learning theory of the active learning-based EMG pattern recognition system, as well as concrete algorithms and experiments. It aims at solving key scientific issues such as the feature representations of the movement information in the stochastic EMG signals, the information fusion method between the reinforcemental learning agent and the EMG pattern classifier, and the theoretical ensurance of the effectiveness and convergence of the learning method. It can provide intelligent biomechatronics research with new design principles and methods. It is expected that this project can accelerate the industrialization of the novel dexterous prosthetic hand, promote the techniques of the rehabilitation equipments in our country, and offer new design concepts in the intelligent biomechatronics systems.

英文关键词： dexterous prosthesis；myoelectric signal control；machine learning；pattern recognition；