A fuzzy expert system for the diagnosis of equipment failure

In reliability engineering, Fault Tree Analysis(FTA) is a powerful assessment method which can analyze a system fault both qualitatively and quantitatively. In qualitative analysis, it helps us to describe an accident model and interpret the relations between the top system-related accident and the basic events which cause the system to fail. In quantitative analysis, given all the original accidents' probabilities, we can evaluate the top accident's probability and its other parameters' values using the fault tree. Since the effect of a cause on an event is usually fuzzy and it is difficult to identify a quantitative value to measure the degree of effect, fuzzy sets theory is introduced into FTA. The fuzzy FTA can handle: 1) uncertainty in failure probability, 2) linguistic descriptions for the fuzzy states of the basic events such as seldom or frequent and 3) linguistic descriptions for the effect of the basic events such as normal or abnormal. In this paper, a fuzzy expert system which can be used to assess the reliability of industrial equipment using fuzzy FTA technique is described. The system is built by LPA-PROLOG and it runs on the Windows platform. The graphical user interface allows the user to build the fault tree from scratch through interactive dialogs. The user can specify either probability values or fuzzy linguistic descriptions. Trapezoidal fuzzy sets are used in developing the verbal statements and in evaluating fault trees. A diagram of the fault tree will be shown at the end of the interactive input process. The system employs a fuzzy importance index to assess the contributions of basic events and the user can use this information to improve the reliability of the system. A case study on an industrial compressor has been conducted and comparisons with results obtained by previous researchers shown the system to be accurate.