Learning about a control domain through expert systems

The basis theme of the paper is that tasks which entail dealing with significant amounts of complexity and uncertainty require for their solution a symbiotic partnership of man and machine. In teaching control engineering, this viewpoint leads to a concern with developing knowledge based support systems (KBSS) rather than autonomous expert systems. These systems must be integrated with a human, so that they are flexible enough not to constrain the human´s powerful reasoning methods and at the same time are capable of teaching control engineers about new design methods. MAID is a prototype system, which has been implemented in the domain of multivariable frequency domain control systems design. An important function of MAID is to teach control engineers new design methods. The designer learns how to generate controllers that manipulate the systems frequency characteristics into a form satisfying the design objectives. The ideas outlined have been used to develop a KBSS for the Multivariable Control toolbook built on top of the MATLAB environment