A new active vibration control architecture using CMAC neural networks

It has been shown previously that CMAC is an effective tool for vibration control. The CMAC network was used in a feedback control structure to produce the signal required to actively cancel the vibration source. This paper offers two significant extensions, which make the CMAC controller method applicable to a wider range of practical problems. First, a new weight update procedure that separates the training cycle from the control cycle is proposed to deal with the phase shift inherent in the plant. Second, the new approach does not require direct measurements of the vibration source. The new vibration control scheme was tested on a submarine simulation model. Results indicate CMAC is an effective tool for this vibration control problem.