The System Design And Recognition Of Electromyography Prosthetic Hand Based On Fuzzy Neural Network

In recent years,our disabled population has gradually increased,especially the number of physically disabled persons increased the most obvious,and so we need study the new generation prosthetic hand with intelligent,making prosthetic hand with higher adaptive capacity.It not only can improve the quality of life of disabled people,but also can produce great economic benefit and social benefit.s EMG(Surface electromyography)signal is a reliable source of the control.Through the identification of the body s EMG signal,it can be used for muscle movement,muscle damage diagnosis,medical rehabilitation,sports and humancomputer interaction,etc.Therefore,the research pattern recognition of the body s EMG signal has a great important scientific value and practical value.This paper will design an acquisition and recognition system of EMG prosthetic hand based on Fuzzy Neural Network,obtain the high quality EMG signal,and make gestures planning and identification,improving the recognition accuracy of EMG prosthetic hand.Firstly,this paper conducts a detailed review for the development status and trend of EMG prosthetic hand at home and abroad,and sums up the main problems existed in the present.Then it analyses the generating mechanism and characteristic of EMG signal,and s EMG signal is affected by several interference forms in the collecting process,and then devises the EMG signal acquisition system.The acquisition system includes the conditioning circuit of EMG signal and data acquisition system software.We respectively carry on simulation test for the trapped wave and filter of the electrode circuit using EWB simulation software,verifying the correctness of the circuit and adjusting the parameters of the capacitance and resistance.Secondly,we preprocess the collected EMG signal,and design a blind analysis method to carry on the filtering of the whole cycle and sliding the whole cycle,and using the correlation analysis makes noise reduction processing for power frequency 50 Hz and its harmonic.We conduct feature extraction for six kinds of hand movements by using the average absolute value,root mean square method,wave length and variance of the time-domain analysis method,making comparison and analysis.Finally,this paper combines fuzzy inference and the advantages of neural network learning adjustment ability,and designs a kind of dynamic fuzzy neural network and its algorithm to learn and establish model for hand action pattern.The feature vector of EMG signal extracted is send into the fuzzy neural network classifier,identifying the palm back,palm front,palm flap,palm upturn,palm fist and palm stretch of the six hand movements.The experiment shows that the fuzzy neural network based on EMG signal has the advantages of high efficiency on a variety of hand movement pattern recognition,and the using different methods of feature to extract recognition rate are also different.