Stone impact damage to automotive paint finishes-a neural net analysis of electrochemical impedance data

Automotive car bodies are subject to impact by stones either lofted from tires or launched by other passing vehicles. Impact can result either in physical loss of paint and the possibility of failure at the metal/phosphate-polymer interface. A neural network (NN) analysis of electrochemical impedance data is presented. It is shown that electromechanical impedance spectroscopy (EIS) is a very sensitive post impact diagnostic probe to detect delamination at the metal-polymer boundary. Considering the noisy quality of data, the learning of the NN is good. It is shown that the NN is able to make predictions that are in agreement with independent experimental observations. Based on this preliminary work the future use of the NN as a predictive tool will rely on a comprehensive data set obtained under rigorous experimental conditions using stone projectiles, alternate treatments of impedance data, and also taking into account parameters such as stone shape, mass, and density