Design of a Fault‐Tolerant System Based on Knowledge‐Engineering Approach and Its Application to a Digital Control System

It has been known that independency between processings for replicated modules is the most important factor in a replicated redundant fault‐tolerant system. To reduce the effects of simultaneous failures and/or simultaneous errors including errors caused by the input itself, a new design method of a highly reliable digital control system based on a knowledge‐engineering approach is proposed. The input signals are sampled at different sampling points and processed separately. If the errors in the processed results are detected using the knowledge of erroneous states and dynamic characteristics of the control system, then the internal states in the erroneous module are recovered immediately by using correct states in the other module. Since the input signal sampling and its processing are almost independent, an ultrahigh reliability can be achieved. The usefulness of the proposed method is confirmed by reliability analysis based on Markov graphs and experimental measurement.