Research On Cooperative Control Of Mechanical Arm Based On Biofunctional Signal

With the development of technology,more and more industrial robots begin to work in the factory instead of manual work to complete high repeatability jobs,and human-computer cooperation mode manipulator has become an important development direction in future industrial field.Surface electromyographic signal is one of the easiest biofunctional signal to collect and process.This paper focuses on human-computer interaction control of industrial cooperative manipulator,propose an analysis method of upper limb movement based on s EMG,and provide theoretical and technical basis for improving the universality and interaction of industrial cooperative robot.First of all,an analysis is made based on the process of human-computer cooperation in industrial production and five groups of human-computer cooperation actions are planned to ensure the mapping relationship between operator and manipulator.The relevant upper limbs of the operator are analyzed to confirm the best position to collect their s EMG signal.After the signals of participants are collected of the Four Channels,they provide data basis for subsequent identification and analysis work.Secondly,the movement translation method is used to detect the starting and ending points of the collected four-channel s EMG signals in order to confirm the starting and ending points of different modes of action,and divide the effective s EMG signals of five modes of action.Two wavelet threshold methods are used to denoise the s EMG signals.By comparing the denoising waveform and denoising index,it is concluded that the layered threshold has better denoising effect on the upper limbs' s EMG signals.From the perspectives of time domain,frequency domain,and time-frequency domain,the overlapping sliding window method is used to extract the features of the pretreated s EMG signal.Through the comparative analysis of the clustering effect of the eigenvalues,the singular value of the two-layer wavelet coefficient with the best clustering effect is used as the feature vector of the upper limb s EMG signals for subsequent classification and recognition.Finally,the extracted eigenvalues are adopted as the input of the classification model by three methods of the support vector machine algorithm,BP neural network,and LSTM neural network to classify and recognize the five upper limbs action patterns.The recognition accuracy of the algorithm is compared and analyzed.According to the experimental result,it shows that the LSTM neural network algorithm has the highest recognition accuracy for human-computer cooperative control action.The accuracy and real-time performance of the analysis method are verified by using a 7-DOF robot arm,verifying the feasibility of the study in this paper.