Relation between disturbance observer and model error compensator

This paper clarifies the relationship between the disturbance observer and the model error compensator. The model error compensator proposed by Okajima et al. (2013) compensates for the error between the output of the nominal plant model and the measurement outputs of the actual plant. On the other hand, the disturbance observer was first proposed to cancel out the effect of the disturbance by using the inverse model of the plant. Okajima et al. (2013) claim that the model error compensator differs from the disturbance observer because it does not use an inverse model, which is an advantage of the model error compensator. However, Ohnishi et al. (1996) proposed the Luenbergar observer-based disturbance observer, which does not use an explicit inverse model. Therefore, this paper shows that the disturbance observer is a type of model error compensator that uses a nominal model. Furthermore, it is shown that the observer gain of the disturbance observer determines the parameters of the model error compensator.