抄録
We review several Just-In-Time methods for data-driven control, in particular which have been developed by the author, and discuss the potential for expantion of this research field. In the Just-In-Time methods, an estimated function is approximated locally with respect to a query to use the training data generated by the estimated function. The training data are stored in a database and the only most related data to the query are searched and extracted for the approximation. In this article, we introduce PID controller tuning, predivtive control, and fault detection based on the Just-In-Time concept.
