Development of a neural network based virtual sensor for automatic transmission slip

Neural networks (NN) are a systematic approach for modeling nonlinear systems. While this technology possess strong appeal and unfulfilled potential, its mainstream integration into many industrial applications has been slower than originally expected. To bridge the gap between NN theory and practice it is essential to evaluate NN performance in practical applications where well-defined design goals exist. Virtual sensor design is an ideal application for a NN approach. Here, the NN would be used as an estimator to potentially replace a hardware sensor. This paper details the development of a NN based virtual sensor for automobile transmission slip estimation.