ANALYSIS OF EMG SIGNAL AND ITS APPLICATION IN PROSTHESIS CONTROL

Abstract

The rehabilitation of an amputee requires the completion of several tasks. The missing body part must be replaced by some artificial meehanical counterpart which should satisfy the requirements of shape, size, weight, strength and cosmetic appearance. A proper communication must be established between the subject and the prosthesisto have proper control and co-ordination of various functions of the artificial limb. The control part is of utmost importance because the acceptance of the prosthesis depends very much upon the type of the control strategy. The control strategy should not require rigorous training, more man-prosthesis interface, and much physical or mental strain. The work embodied in the thesis is towards the completion of the control task. The EMC is used for control of limb prosthesis systems. The EMC signal and its source dynamics are analyzed for developing the prosthesis control systems. A mathematical model for describing the spectrum of EMC signal has been developed. Motor unit inside the source muscle is considered as the basic functional unit for the generation of EMC signal. The placement of the electrodes on the skin above the source muscle is critically examined. The effect of conduction velocity, frequency of the signal generation, number of the motor units inside the source muscle are discussed. The model gives a picture of the EMC signal picked-up with the help of surface electrodes for the prosthesis control applications. Single function and multifunction control systems are developed for the limb prosthesis control purpose. An EMC processing unit is developed for picking-up the EMG signal from the source muscle. The unit has high input impedance,1arge c ommon-mode-re jection ratio, high signal to noise ratio and low thermal sensitivity. The power line interference is fully suppressed. The EMG processing unit consists of a preamplifier, high gain amplifier, low-pass filter, 50 Hz rejection filter and an emitter follower. This unit can be used for picking-up the EMG signal for any type or level of prosthesis control. Two different types of EMC- operated prosthesis control systems have been developed for controlling the operation of knee—joint In an above—knee—prosthesis using one EMC processing unit and one feedback unit. The EMG processing unit picks and processes EMC signal. The feed back unit picks-up the signal from the prosthesis about its state. The feedback unit consists of a pressure-transducer, d.c. amplifier, and a level detector. The prosthesis control

systems operate on a LOGIC which has been developed after study of the locomotion cycle of natural leg for normal operations such as walking, climbing the stairs, etc. The first control system consists of an EMC processing unit, feedback-unit, fullwave rectifier, level detector, logic circuit and a driver amplifier. Second system consists 01 an EMC processing unit, feedback unit, pulse shaper, bistable circuit, logic gate and two driver systems. The systems are highly reliable In operation, need less man-prosthesis inter face. The second type of control system has one advantage that the source muscle has to be excited for a very short time and the power consumption is very less. The prosthesis control systems have been successfully tested for controlling the knee joint operation. To solve the problem of multifunotional prosthesis control, the author has developed two methods. The method will not put mental or physical strain on the subjects. in one of the methods, the control of the prosthesis control system via EMG signal using autocovariance(£07) parameter estimation has been, given. The method presents reliable separation and distinction of the EMG signal corresponding to various limb functions. The control strategy is applicable for any type or level of prosthesis control i.e. for both upper and lower extremities.. The experimentally recorded EMC data from the source muscle (biceps muscle) for the various

limb functions is analyzed and the validity of the new method is established. The repeatability of the ACV uarameter for the limb function over an extended period of time is shown. The realization problem of the prosthesis control system has been discussed. The second method is presented for multi-functional prosthesis control using finite state automata theory. The control method is generalized type and is applicable for any type or level of prosthesis control. In this method, the EMG signal is required about the state of the natural remnant part of the limb. The other control signals are taken from the prosthesis itself. The validity of the method is established by designing a finite state controller for an above-lmee-prosthesis. The control system consists of sequential circuits, threshold gates and cybernetic actuators. The system is autonomous in operation, stable at all instances due to feedback signals, has less man-prosthesis interface, uses optimum number of components and has high reliability. The method is experimentally verified for two input signals i.e. one EMG signal and one feedback signal for knee-joint operations. Further, the EMG signal source dynamics is quanti- . fled using automata theory. The model gives an idea of EMG generation urocess which is helpful while developing the EMG

operated prosthesis controllers. Also it is useful in research and diagnostic purposes; The motoneuron cell body is modelled as weighted threshold gate and the bundle of muscle fibers as sequential machine. The sequential machine model of motor unit is justified as it provides memory facilities to incorporate the events of the motor unit ove: an extended period of time. The spatial summation of the fiber's potentials is also discussed.