In feedforward active noise control problems, it is always necessary to optimise both the physical system and the electronic controller. Optimisation of the physical system is principally concerned with determining the optimum location of the control sources and error sensors. Here, genetic algorithms are investigated for this purpose and modifications to the standard algorithm that are necessary for this application are discussed. Optimising the electronic controller is principally concerned with finding the optimal control filter weights that will produce the most noise reduction when the reference signal is filtered and then input to the control sources. Normally, gradient descent algorithms are used for this purpose. However, for nonlinear systems, such as control sources (loudspeakers or shakers) with significant harmonic distortion, the gradient descent algorithm is unsatisfactory. Here a genetic algorithm is developed specifically for control filter weight optimisation. It is able to handle nonlinear filters and requires no cancellation path identification. The disadvantage is that it is relatively slow to converge.

Inspec keywords: vibration control; nonlinear filters; nonlinear control systems; acoustic signal processing; feedforward; genetic algorithms; active noise control

Other keywords: nonlinear system; nonlinear filters; noise reduction; error sensor; control source location determination; control filter weight optimisation; feedforward active noise control problem; reference signal filtering; ASVC system optimisation; electronic controller; active sound and vibration control; genetic algorithm; physical system optimisation

Subjects: Acoustic noise, its effects and control