Research on the reliability of automotive sensors based on minimum relative information entropy

Abstract

Based on the principle of minimum relative information entropy, a new compound assignment method is proposed to establish an automotive sensor reliability evaluation model as an example. In order to improve the ease of use, a BP network is constructed with a single indicator as the collection sample and the diagnosis result as the output sample. It is verified by MATLAB simulation that the average relative error of prediction is 0.64% and the maximum relative error is 1.61%, which indicates that the model can give evaluation results quickly within a reasonable range and has certain application value for automotive sensor reliability evaluation.