抄録
In this paper, characteristics of a couple of model following controls is considered. One is ordinary control to follow the states of the model (model following control) and the other one is the control to follow the behaviour of the model (model-parameter following control) . To indicate the differences, two sliding mode controllers are introduced. As a result, model-parameter following sliding mode controller, controller based on a disturbance observer, and learning controller belong to the latter. By this classification, it is shown that the controller based on a disturbance observer rejects the disturbance approximately and rejects it completely if and only if the disturbance is constant. Moreover a new gain to improve the performance of the controller based on a disturbance observer is indicated. Here the performance is a measure of the time it takes to reject the disturbance (restrict the system onto a hyper plane) . To clarify the comparison between the new and the original gains, an LQ evaluative function is offered. Some numerical simulation results are presented.
