【作者】 陈斌；

【导师】 刘继忠； 王维斌；

【作者基本信息】 南昌大学 ， 机械工程（专业学位）， 2021， 硕士

【摘要】 表面肌电信号是人体运动时肌肉产生的生物电信号,反应肌肉动作的状态。表面肌电信号由于易采集,提取方便等特点被广泛应用在康复医疗、运动医学、智能机器人等领域。本文基于表面肌电信号的手势动作识别研究,根据识别结果去控制护理床完成相应的动作,不仅可以满足肢体残疾病人对智能护理床自主操控的迫切需求,而且在康复医学领域提供更为可靠和人性化的辅助康复设备。因此,具有非常重要的研究意义和应用价值,本文主要工作如下:(1)对表面肌电信号的产生机理和特点进行了介绍,确定了采集表面肌电信号的方式,即非侵入式采集,通过表面电极贴片采集实验者的肌肉皮肤表面肌电信号。(2)实验共设计了4种手势动作,通过窗口分析法对表面肌电信号的起始位置进行分割,得到手部活动段信号。对分割得到的表面肌电信号采用五阶巴特沃斯带通滤波器滤波,再对滤波后的表面肌电信号采用CEEMDAN-小波阈值法降噪处理。最后对降噪后的表面肌电信号进行时域和频域特征提取,采用主成分分析法PCA对提取的肌电信号进行特征降维处理,对下文的手势分类的分类精度提升起到了重要的作用。(3)采用近邻算法,BP神经网络算法和支持向量机算法,对不同手势动作模式的肌电特征进行分类识别,得出支持向量机模型分类效果最好,识别率为96.43%。最终选择了支持向量机算法作为手势动作识别的应用研究。(4)设计了手势动作识别与护理床控制实验。对5名受试者开展基于表面肌电信号的护理床控制实验,通过四种手势动作去控制护理床运动并且对准确率与实时性进行评估。实验结果表明,在手势识别的动作下,护理床可以完成运动控制,可以帮助偏瘫患者完成自主康复训练。更多还原

【Abstract】 Surface EMG(s EMG)is the bioelectric signal produced by the muscle during human movement,which reflects the state of muscle movement.Surface EMG signal is widely used in rehabilitation medicine,sports medicine,intelligent robot and other fields due to its easy acquisition and convenience.In this paper,the gesture recognition research based on surface EMG signals is studied.According to the recognition results,the nursing bed can be controlled to complete the corresponding movements,which can not only meet the urgent needs of physically disabled patients for the autonomous control of intelligent nursing bed,but also provide more reliable and humanized auxiliary rehabilitation equipment in the field of rehabilitation medicine.Therefore,it has very important research significance and application value,The main work of this paper is as follows:(1)The generation mechanism and characteristics of surface EMG signals are introduced,and the way to collect surface EMG signals is determined,which is noninvasive collection,and the muscle and skin surface EMG signals of the experimenter are collected by surface electrode patch.(2)A total of four kinds of gestures are designed in the experiment,and the initial position of the surface EMG signals is segmenting by window analysis method to obtain the hand movement signals.The segmentation surface EMG signals are filtered by fifth order Butterworth bandpass filter,and then the filtered surface EMG signals are denoised by CEEMDAN-wavelet threshold method.Finally,the surface EMG signals after noise reduction are extracted in time domain and frequency domain,and the principal component analysis method(PCA)is used to carry out feature dimensionreduction processing on the extracted EMG signals,which played an important role in improving the classification accuracy of the following gesture classification.(3)The nearest neighbor algorithm,BP neural network algorithm and support vector machine(SVM)algorithm are used to classify and recognize EMG features of different gesture movement patterns.The results showed that the SVM model had the best classification effect,and the recognition rate is 96.43%.Finally,the support vector machine algorithm is chosen as the application research of gesture recognition.(4)The experiment of gesture recognition and nursing bed control is designed.The experiment of nursing bed control based on surface EMG signals is carried out on5 subjects.Four kinds of gestures are used to control the movement of nursing bed,and the accuracy and real-time performance are evaluated.The experimental results show that the nursing bed can complete the motion control under the gesture recognition,and can help the hemiplegia patients complete the autonomous rehabilitation training.更多还原