Identification of Unknown Inputs in Nonlinear Systems Based on Interval Methods

This paper considers the problem of estimating (identifying) the magnitude of unknown inputs in engineering systems described by nonlinear dynamic models with measurement noises, which is necessary to compensate for the effect of such inputs. The solution of the problem is based on minimal-order interval observers, and their explicit relations are obtained. Such an observer and a differentiator are used to derive relations for an interval estimate of the magnitude of unknown inputs. The theoretical results are illustrated by a practical example of estimating the magnitude of unknown inputs on an electric drive of a manipulator robot. The simulations in MATLAB demonstrate the correctness of the theoretical results. The above relations can be used to design fault-tolerant systems.