Modelling and FDI of Dynamic Discrete Time Systems Using a MLP with a New Sigmoidal Activation Function

Abstract

In this paper we investigate the use of the multi-layer perceptron (MLP) for system modelling. A new sigmoidal activation function is introduced and the study is focused at the utilization of this function on a MLP that performs modelling of dynamic, discrete time systems. The role of the activation function in the training process is investigated analytically, and it is proven that the shape of the activation function and it's derivative can affect the training outcome. The method is simulated at a well known benchmark, namely the three tank system, and is incorporated in a Fault Detection and Identification (FDI) method, also applied and simulated at the three tank system. Finally, a comparison is made with an approach that utilizes local model neural networks for system modeling.