Probabilistic reasoning in intelligent systems: networks of plausible inference
Probabilistic reasoning in intelligent systems: networks of plausible inference
Abductive inference models for diagnostic problem-solving
Abductive inference models for diagnostic problem-solving
System Test and Diagnosis
| Hi-index | 4.10 |
As systems grow in complexity, the need for reliable testing and diagnosis grows accordingly. The design of complex systems has been facilitated by CAD/CAE tools. Unfortunately, test engineering tools have not kept pace with design tools, and test engineers are having difficulty developing reliable procedures to satisfy the test requirements of modern systems. When testing a complex system, what is the proper response to unexpected and conflicting test results? Should the results be scrapped and the tests rerun? Should the system be replaced or redesigned? Should the test procedures be redeveloped? Determining the best answers to these questions is not easy. In fact, each of these options might be more drastic than necessary for handling conflict in a meaningful way. In this article, we discuss two approaches to system diagnosis that apply several tests and interpret the results based on an underlying model of the system being tested. These tests are used to determine if the system is functioning properly and, if not, to explain the faulty performance. By examining potential sources of conflict and the way conflict might become manifest in a reasoning system, we developed an approach to extend diagnosis to handle the conflict and draw more reliable conclusions.