摘要:AbstractThis note proposes a neural network-based methodology to compute the equivalent control that enforces the robust stabilisation of a class of discrete-time nonlinear systems, which are subject to a wide variety of disturbances. A step-back disturbance observer is firstly introduced, such that, the apparent disturbance is related to the first-order variation of the unknown term. The effect of the apparent disturbance is compensated by deeming on the universal approximation of radial basis function networks, whose parameters are computed by considering the gradient descent of a suitable cost function, such that, the system evolves with a negligible effect of the disturbance. A representative simulation highlights the reliability of the proposed approach.
关键词:KeywordsNeural networksDiscrete-time systemsAdaptive controlRobust control