Performance monitoring in nonlinear adaptive noise cancellation

Correlation based model validity tests are introduced to monitor the operation of nonlinear adaptive noise cancellation filters and to detect if the filters are operating correctly or incorrectly. The tests are derived for a NARMAX (Nonlinear Auto Regressive Moving Average model with eXogenous inputs) filter design based on a SubOptimal Least Sqaures (SOLS) estimation algorithm. Simulation studies are included to illustrate the performance of the new tests.