A frequency domain method is developed for statistical multivariable plant set estimation. The estimation of a plant “set” rather than a point estimate is required to support many methods of modern robust control design. Results for the multivariable case extend earlier results developed for the single-input single-output (SISO) case. The approach here is based on using multisinusoidal input designs, and acquiring multivariable data from a sequence of single-input multiple-output (SIMO) experiments. Estimators of the plant, output noise, and uncertainty are defined by using the Discrete Fourier Transform (DFT), and statistical properties of the estimators are presented. These properties lead to a precise characterization of the plant set to a specified statistical confidence, e.g., (1-α)-100%.
Read full abstract