FLC Design with Multiple Control Objectives and Application to Overhead Crane Control

Abstract

The fuzzy logic control can be viewed, in a certain sense, as a means to model a human operator in human-in-loop system[l]. However, when the system has many control objectives to satisfy and/or many variables to refer to, it is often very difficult to make the rule base for the FLC(Fuzzy Logic Controller). The control of the overhead crane can be an example(See Figure 1)[2][7][8][9]. The control objectives of the overhead crane are two fold; one is positioning the trolley as fast as possible and the other is reducing the swing of the load as small as possible. The operators of the overhead crane in the Pohang Iron & Steel Company in Korea are controlling the crane so as to satisfy both of the control objectives as much as possible. However, when we had interviews with the operators to make the rule base for the FLC, we found that the operators described their behaviors in terms of fuzzy If-Then rules referring to only one of the two control objectives. That is, they could explain their behaviors of positioning the trolley only without referring to reducing the swing and could explain their behaviors of reducing the swing only without considering the positioning of the trolley. Nevertheless, the operators were claiming that they can exert both actions of positioning the trolley and reducing the swing at the same time. For this experience, we have found that, when there are uncertainties in making the rule base for controlling a plant with many control objectives to satisfy, we may obtain separate groups of the rules satisfying partial control objectives each instead of the rules satisfying all the control objectives simultaneously.