An adaptive feedback testing system (AFTS) is presented which includes self-testing and self-modifying functions, in order to enhance the reliability of knowledge-based computer-aided-design software programs. A rule-based expert solvers operating in an uncertain environment can select a plausible solver that results in an error-prone solution. The automatic test system evaluates the candidate solver so that the correct decision is made. Using IDSCA, an expert system built by the authors, to direct the controller's action in a multiloop control system, the function of the AFTS is demonstrated.
Relation:
IEEE Proceedings of the National Aerospace and Electronics Conference 1988, Pages 414-420
