A real time surface electromyography signal driven prosthetic hand model using PID controlled DC motor

• Published in: Biomedical engineering letters Vol. 6; no. 4; pp. 276 - 286
• Main Authors: Raj, Retheep, Ramakrishna, Rejith, Sivanandan, Karingannikalm Subramanian
• Format: Journal Article
• Language: English
• Published: Korea The Korean Society of Medical and Biological Engineering 01.11.2016; Springer Nature B.V
• Subjects: Angular velocity; Biological and Medical Physics; Biomedical Engineering and Bioengineering; Biomedicine; Biophysics; Control methods; Controllers; D C motors; Electric motors; Electromyography; Engineering; Forearm; Human performance; Kinematics; Medical and Radiation Physics; Motor task performance; Muscles; Neural networks; Original Article; Parameter estimation; Proportional integral derivative; Prostheses; Real time; Regression; Root-mean-square errors; Simulation; Statistical analysis; Time domain analysis; Velocity; MLPNN; IEMG; sEMG; ZC; Forearm kinematics; NARX structure; PID controlled prosthetic hand
• ISSN: 2093-9868; 2093-985X
• DOI: 10.1007/s13534-016-0240-4

Purpose A surface electromyography (sEMG) driven proportional-integral-derivative (PID) control method is proposed to control the prosthetic hand model according to human intentions in real time. Methods The sEMG signals are acquired from the biceps and triceps brachii muscles of the human hand from 30 able bodied subjects. Two time domain features, integrated EMG (IEMG) and number of zero crossing (ZC) are extracted from the sEMG signals and these features are used for the estimation of human forearm kinematics. The estimation of human forearm kinematics is achieved by multi layered perceptron neural network (MLPNN) model based on nonlinear auto regressive with exogenous (NARX) inputs. The estimated human kinematics are utilized to control a direct current (DC) motor based prosthetic hand model using PID controller. The controller parameters are tuned manually to obtain the best possible results. Results It is observed that the IEMG and ZC varies with change in angular displacement and also with change in angular velocity. The performance of estimation and control is evaluated using two statistical parameters, root mean square error (RMSE) and regression value. The RMSE and regression value obtained during estimation of angular displacement is 5.89 and 0.97 and the corresponding value obtained for the estimation of angular velocity is 18.91 and 0.80. The RMSE and regression value obtained during control of angular displacement is 18.9096 and 0.9456 and the corresponding value obtained for the control of angular velocity is 27.91 and 0.68. Conclusions Experimental results confirm that the estimation using MLPNN and PID controlled prosthetic DC motor hand model performs well. The proposed method is simple in design and can be implemented in a human being with fewer costs.