An Optimization Approach to Fuzzy Diagnosis: Oil Analysis Application

Abstract

This paper discusses a knowledge-base encoding methodology for diagnostic tasks. It transform "expert"-provided rules into algebraic expressions so inference of the "possible" disorders is carried out via associated constrained optimisation problems. In this way, the need of conventional fuzzy inference systems or "uncertain"-logic schemes is no longer present in the particular setting in this paper. An oil-analysis diagnosis case study is presented as an application example, with actual experimental data. The problem is solved by efficient linear programming tools, in principle able to cope with large-scale problems. The only software used was Mathematica reg 5.2.