Robust process identification from relay tests in the presence of nonzero initial conditions and disturbance

In this paper, a new method is presented for process identification from relay tests. The method can yield a full process model in the sense of a complete transfer function with delay or a complete frequency response. Unlike our early work that was based on the fast Fourier transform (FFT), where zero initial conditions and no disturbance are assumed, the present method can work under nonzero initial conditions and static disturbance. It adopts the integral technique to devise the algorithm. The resulting integral-based estimation shows robustness against noise in output measurements but requires no filtering of noisy data. Its effectiveness is demonstrated through simulation.