Although the idea of utilizing expert systems in maintenance held early promise, the use of rulebased programming has led to practical problems in implementation.
For example, XCON, an expert system developed by Digital Equipment Corp.
for product configuration, has more than 10,000 rules.
Issues such as maintainability of the knowledge base, testability of the program, and reliability of the advice have limited the practical use of most expert
systems in maintenance.
Other approaches, such as constraint-based reasoning, are being developed as
alternatives to rule-based systems) 5 Also, the reconsideration of Bayesian theory to support probabilistic reasoning and maintenance diagnostics is being reexamined.