Application of multi-layered feedforward neural networks in digital vibration control

Summary form only given, as follows. The authors discuss the application of multi-layered feedforward networks (MFNs), with the delta bar delta backpropagation learning rule, to the problem of digital vibration control of mechanical systems. The results of conventional control were compared with the results of a controller which uses a trained MFN. The results clearly show the superior performance of the neural network control. This enhancement of performance was attributed to the ability of a neural network to produce a better sampling period phase delay compensation and a reduction and filtering of the higher frequency noise. In conventional implementation of digital control the noise is the result of a phenomenon referred to as controller spillover