Failure analysis is a complex task, requiring consick rable knowledge and skill, some of which might outside the experience of the engineer concerned. To help the engineer, a case-based reasoning tool (Failure Analysis Diagnostic Expert System) is being developed, using a systemic approach to failure diagnostic problems, which not only interactively identifies the failure modes and the critical factors the design, processing, and end use which cause failures to originate but also suggests methods improve the reliability of the products. This approach applies a Windows artificial intelligent case-has reasoning technique to generic failure diagnostic problems. Currently, the research is based around the elicitation of failure analysis knowledge and the smlcturing and planning of this knowledge so that the expert system behaves in an intelligent and responsive manner. Many problems have arisen, such understanding what information is required to identif the failure mode and in designing a consistent a economical set of questions that are integrated into an efficient questioning strategy. For the behavior of the system to be intelligent and responsive, research into interface design and the understanding human cognitive models for failure analysis decision-making is proposed. The development of structure models will show which factors are critical in a given situation and aid failure analysis diagnosis.
Keywords: failure analysis, failure diagnostic, case-based reasoning, rule-based reasoning, expert systems, materials failure, modelling, computer-aided learning.