A Novel Fault Diagnosis Framework of Computing with Words Based on Lattice-Valued Fuzzy Automata

Abstract

Fault diagnosis is crucial for complex man-made system. The discrete event system approach is an efficient method for this problem. However, the states, events and transition functions in the previous discussions were assumed to have crisp values. This prevents the use of the existing approach for some other important fields when uncertain problems need to deal with, such as uncertain states, fuzzy events. Existing studies on classical fuzzy automata for fuzzy discrete event system fault diagnosis cannot solve all the problems encountered, such as dealing with fuzzy event string, fuzzy subset of events and so on. In this paper, we propose a more general fault diagnosis framework for fuzzy discrete event system based on the lattice-valued fuzzy automata. In this framework, the system is modeled as a fuzzy discrete event system based on lattice-valued fuzzy automata. Based on this model, a fault diagnoser for system is constructed, which is also a lattice-valued fuzzy automaton that can be used to detect system faults online. The diagnosis model of computing with words is also established by extending its state transition functions on fuzzy subset of observable events. The heating, ventilating and air condition system simulation experiment is carried out, and the results show that the lattice-valued fuzzy diagnoser is a promising way for fault diagnosis of system with uncertain problems.