Prediction for Dynamic Systems Using Impulse Response Representation

A predictor for a dynamic system that maps past inputs and outputs, along with future inputs, to future outputs is presented. The predictor models noise and disturbances as Gaussian random variables, and a maximum likelihood and minimum variance solution are provided. The novel element of this predictor is that it is defined directly in terms of the system impulse response, eliminating the need to find a state-space or transfer function realization.