摘要
作为下肢康复外骨骼机器人的核心,人体意图识别一直是国内外最热门的研究课题。文章针对人体意图识别中下肢关节角度预测精度较低的问题,提出了一种基于表面肌电信号的多特征值融合的遗传算法(Genetic Algorithm,GA)优化BP神经网络对下肢关节角度进行预测。利用表面肌电采集设备采集人体下肢表面肌电信号,对信号进行处理并提取多个特征值,将提取的多个特征值作为神经网络的输入,同时将采集到的人体髋、膝、踝关节角度作为输出。实验结果显示,与传统BP神经网络结果相比,遗传算法优化BP神经网络具有更高的预测精度,且预测结果更加稳定。
As the core of lower limb rehabilitation exoskeleton robots,human intent recognition has always been the hottest research topic at home and abroad.In the paper,aiming at the problem of low accuracy of lower limb joint angle prediction in human intent recognition,a genetic algorithm(GA)based on multi-feature value fusion of surface EMG signals is proposed to optimize the BP neural network to predict the angle of lower limb joint.The surface EMG acquisition device is used to collect the surface EMG signal of the human lower limb,process the signal and extract multiple feature values,take the extracted multiple feature values as the input of the neural network,and take the collected human hip,knee and ankle angles as output.The experimental results show that compared with the traditional BP neural network results,the genetic algorithm optimizes the BP neural network with higher prediction accuracy and more stable prediction results.
作者
邓福铃
DENG Fuling(Chongqing Jiaotong University,Chongqing 400060,China)
出处
《传感器世界》
2023年第6期16-20,37,共6页
Sensor World
