Nonlinear recursive identification of electrically stimulated muscle

A recursive constrained least-squares (RCLS) identification method is applied to the identification of electrically stimulated quadriceps muscles in paraplegic human subjects, using percutaneous intramuscular electrodes. The nonlinear time-varying steady-state torque vs. pulsewidth recruitment characteristic of the electrode and muscle is identified simultaneously with the time-varying input/output muscle response dynamics, using a Hammerstein-type model which has been modified to include the recruitment threshold (deadband). Knowledge of the recruitment curve´s shape is translated into linear equality constraints on the identified parameters, which are enforced by the RCLS algorithm. It is found that the enforcement of parameter constraints results in reduced parameter variance and enhanced future output prediction capability