Design of estimator with disturbance model under parameter uncertainty and its application to autopilot

We propose a practical design method of an estimator based on observer which includes a disturbance model when uncertainties exist in nominal parameters. The procedure first designs an estimator without the disturbance model, and designs an estimator with the disturbance model next. The characteristic of the estimated error which originates from parameter uncertainties by the former is investigated, and the characteristic of the disturbance elimination which originates from the disturbance model by the latter is investigated. Consequently, those characteristics of the estimator with the disturbance model were clarified and the disturbance elimination performance of the estimator with the disturbance model improved 2.5 times compared with the case of the estimator without the disturbance model.