Non-parametric identification of homogeneous dynamical systems

The aim of this study is to design a non-parametric identifier for homogeneous systems based on class artificial neural networks with continuous dynamics. identification algorithm developed input-affine uncertain gains and diverse degrees homogeneity. One the main contributions extension universal approximation property systems. Another contribution development differential novel concept networks. adjustment laws weights are obtained from Lyapunov stability analysis taking homogeneity properties system into account. ultimate boundedness origin error demonstrated using persistent excitation condition. effectiveness proposed verified by simulation three-tank model. In example, scheme compared classical ANN identifier, we present statistical such comparison. It shown in simulations that has faster convergence less oscillations.