Learning control systems with relational plants

The control system with a static plant described by the knowledge representation in the form of relations is considered. The learning process consists in using the successive knowledge validation and updating to the determination of the current control decisions. Two approaches and algorithms have been described: for the validation and updating of the knowledge on the plant and on the form of the control. In both cases two versions are presented: the learning process in open-loop and in closed-loop control system. A special case of the relations and simple examples are included.