A nonlinear approach to modeling of electrically stimulated skeletal muscle.

This paper is concerned with the development and analysis of a nonlinear approach to modeling of the contraction of electrically stimulated skeletal muscle. The model structure is based on a network of locally valid linear models which are blended together by a scheduler. Data are from experiments with rabbit tibialis anterior muscles in which the muscles contracted isometrically while being stimulated by supramaximal impulses with randomly varying inter-pulse intervals. The model accounts for nonlinear effects due to variations of the stimulation frequency, such as the "catch-like" effect. It is shown that this modeling technique is suitable for modeling the contraction of muscles with very different characteristics, such as muscle with a majority of fast motor units and muscle with mainly slow motor units. The approach is also suitable as a basis for the design of muscle stimulation controllers. Index Terms-Functional electrical stimulation, local model network, muscle modeling, nonlinear system identification.