FNS Parameter Selection and Upper Limb Characterzation

The design of a controller for functional neuromuscular stimulation requires characterization of the plant, i. e., the neuromuscular system involved. The present paper is concerned with the modeling and identification of the elbow flexion/extension and wrist pronation/supination systems. These movements can be performed successfully by the surface stimulation of the biceps, triceps, and pronator teres. The scatter in the plant parameters between subjects and for a given subject as a function of time, electrode placement, etc., is outlined. The nonlinear muscle gain is approximated by a threshold below which the torque is zero, followed by a linear region. A least squares identification technique using a pseudorandom binary sequence input returned best fit for a third-order model for both the elbow and wrist dynamics. The dependence of the complex poles on the input signal bandwidth and level of coactivation is examined. The differences in parameters between the wrist and elbow systems are calculated. The results are used to establish design requirements for an appropriate controller.