A new model graph called the fuzzy cause-effect digraph (FCDG) model was already proposed in part 1, and its capability to eliminate spurious interpretations attributed to system compensation and inverse responses from backward loops and forward paths is to be demonstrated. In this paper we attempt to develop a new fault diagnosis algorithm based on the fuzzy cause-effect digraph model. This method applies fuzzy reasoning to estimate the states of unmeasured variables, to explain fault propagation paths, and to locate fault origins. In particular, it can obtain the fault origin occurring in the process with single and multiple loops at the early stage of fault. This study uses a CSTR as an example to explicate this diagnosis method and compares the results with those of other methods.
