Recurrent RBF networks for suspension system modeling and wear diagnosis of a damper

From the beginning of the nineties, artificial neural networks began to be used for the identification and control of nonlinear systems. Thus, new types of nets appeared permitting to model the behaviour of dynamical systems. These recurrent networks extend the range of engineering applications and emerged as new powerful architectures. Besides, local models like RBFN (or neuro-fuzzy structures) provide an interesting alternative to the more classical MLP since information is naturally localized within some nodes and not distributed amongst all the nodes. Accordingly, this paper is concerned with the use of recurrent RBFN for modeling and identifying dynamical systems within the framework of an output error approach. Details of the implementation are given for the parametric as well as for the structural optimization. An application to the modeling of a vehicle suspension system is dealt with, intending to provide a wear diagnosis for the damper element. A quarter-vehicle model is developed and the wear phenomena are expressed in the shape of an hysteresis effect. Simulated results are given from data produced by dampers in good and bad conditions.