システム同定によるデータ駆動制御と制御対象の複雑さに関する一考察

抄録

In this study, the purpose is to analysis influences that give plant complexity(model order) control performance in data-driven control via system identification. In the framework of maximum likelihood identification, based on the concept of information criterion, the relation of the control performance and the model order is derived, and it is shown that high order model is tend to fundamentally disturb to make the control performance best.