An Artificial Neural Network for Low-Energy Impact Monitoring

Abstract

An artificial neural network has been developed that can analyze signals from piezoelectric sensors to detect the location and intensity of low-energy impact on a composite panel. The panel has two piezopolymer sensors attached some distance apart from each other. The signals are analyzed by the artificial neural network using a binary pattern associator. The binary pattern associator facilitates memory build-up during learn ing while simplifying signal discrimination during testing. The impact monitoring system, which consists of the artificial neural network and the piezoelectric sensors, has been used successfully to detect the location and height of impact when a small steel ball is dropped onto a composite laminate.