Parameter variations, relative degree, and stable inversion (Q5935480)

The question on continuous dependence of the exact output tracking problem solution (stable inversion process) with respect to parameter variations is studied. Input-output systems with controllable and observable linear parts are considered. It is supposed that the vector relative degree is fixed for each value of the parameter, but it can vary as the parameter varies. The main model of input-output representation of the initial system is the inverse dynamical equations for the control. Linear systems are first studied. With the help of the Fourier series representation, the continuity of input, output, their derivatives and hence of the state is shown. Further, dependence of the impulse response and of an appropriative convolution integral is studied with the change of the parameter. The results obtained are applied for nonlinear systems. A Picard iterative process for nonlinear system approximation is used so that at each step the convolution integral of the linear kernel with a time function is calculated. Thus the problem for nonlinear systems is reduced to a linear one. The theorem proved indicates that the stable inversion process depends continuously on the parameter values. Hence even if the order of differential equations for control can change, the solution is still continuous in the parameter.