Modular identification of the dynamical model of friction force

An alternative approach for the identification of the LuGre dynamical friction model is proposed in this paper. The identification procedure is made using modular neural networks (MNN). The computational complexity of neural identification can be greatly reduced if the whole system is decomposed into several (nonlinear) subsystems. Each local nonlinear model is identified by a neural network (NN). The switching between the neural models is done by a dynamical switcher, also implemented by neural networks, that tracks the different operating points.