Prediction of stability and performance of an active mechanical structure under uncertainty conditions using finite element and neural computation

This paper presents an analysis of stability and performance of a vibration-controlled engineering structural system using the artificial neural network (ANN) method. The application concerned an aluminium panel for which thickness, Young modulus and width were varied. A particular feature was demonstrated which related the prediction of system performance, in terms of vibration attenuation, to the tests required for optimising the neural network. The need for instability information was determinant to accurately describe the effect of each of the variables considered in the complete system.