Genetic algorithm adaptation of non-linear filter structures for active sound and vibration control

This paper examines the use of a genetic algorithm for active sound and vibration control, where the genetic algorithm is used for the adaptation of the filter weights of non-linear filter structures. Use of a genetic algorithm for adaptation of filter weights eliminates the need to know the cancellation path (control source input to error sensor output) transfer function, which is essential knowledge for a conventional gradient descent-type algorithm to remain stable. This enables the use of complex filter structures which would normally not be practical, especially where the cancellation path transfer function is non-linear and requires a non-linear model to ensure stability. Experimental verification of algorithm operation by non-linear vibration control of a beam is presented, using both polynomial and neural network filter structures