Abstract：Objective Multiple characteristics of wavelet features of EEG and EMG, such as time domain, frequency domain, timefrequency
domain and sample entropy, were extracted and analyzed to study the movement recognition of human lower limbs.
Methods The EEG and EMG signals of human lower limb movement modes were acquired from six kinds of motor states including
standing, walking, Up the stairs, down the stairs, up the hill and down the hill. In order to analyze the initial intention and specific
types of lower limb movement, the signals were preprocessed, extracted features and recognized. Results The experimental results
indicated that the signal features extracted from time domain, frequency domain, time-frequency domain and time-complexity can
better represent the characteristics of signals. The average recognition rate of EEG signal classification based on support vector
machines was 93.2%. The EMG signal classification based on extreme gradient boosting had an average recognition rate of 93.6%
for the six lower limb movement modes, which can effectively recognize human lower limb movements. Conclusion The method
proposed in this paper can effectively identify the movement of human lower limbs and accurately identify the movement intention
of human lower limbs, which can provide accurate and safe control strategies for the lower limb exoskeleton robot and improve the
efficiency of the lower limb exoskeleton robot’s walking assistance.
郑长坤，王海贤，顾凌云，张弛，汪丰. 基于脑电和肌电信号的下肢运动意图识别方法[J]. 中国医疗设备, 2021, 36(5): 61-66.
ZHENG Changkun, WANG Haixian, GU Lingyun, ZHANG Chi, WANG Feng. Recognition Method of Lower Limbs Motion Intention Based on EEG and EMG. China Medical Devices, 2021, 36(5): 61-66.
