PERFORMANCE EVALUATION BY QPIT*I.

A real-time process control system requires much more precise response time than a general purpose computer system. The system performance evaluation has three phases; evaluation - diagnosis - improvement, which form a feed-back cycle. Among the three phases, past studies have been focusing mainly on the evaluation phase and many computerized methods were already proposed. On the other hand, the other two phases have been performed based on a trial-and-error method. There have been no standard methods for these two phases. A new method for these phases is introduced here. It provides supports to even non-experts to obtain nearly optimum performance. In this method, diagnosis is done by a simple rule-base system and improvement plans are generated in a heuristic way. An experimental tool called Qpit*i was developed based on this method and is presented in this paper. Qpit*i uses simulation data and produces several improvement plans against spotted bottle-necks in an interactive mode with users.