الفهرس 
introductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Diagnosis is a very complex and important task, where it is for detennining the conditions of patients in the medical field, or finding the root cause of faults in physical systems. In physical systems, such as nuclear process plants, diagnosis involves matching patterns of sensors measurements and process alarms to specific equipment malfunctions and operational faults. Diagnosis can be viewed as the process of linking symptoms to
causes .
The accidents of Three Miles Island (TM!) and Chernobyl have focussed the demand for nuclear power plant safety. This demand has propelled the development and implementation of innovative reactor designs, better safety systems, human factor studies, and stricter safety requirements. Even with these developments and implementations, it is still imprudent to assume that an accident in nuclear power plant can never occur. In normal operation of a nuclear power plant, the plant is controlled by plant trip and control systems. During an abnormal reactor transient, an accurate and true understanding of the state of the reactor is necessary to correctly diagnose and mitigate the transient. A quick and accurate diagnosis is very important to plant safety because relatively simple procedures can usually be implemented to correct the situation. Operators are trained to deal with diagnosis and mitigation of a wide variety of event scenarios. The expert reactor operator can respond to this information from experience, research, learning, or intuition, however, these concepts are not programmable into a mechanistic strategy.
During emergency situations, plant conditions can present a challenge to an operator’s training experience. Information overload, multiple failure, unce~inty, and likelihood of invalid or unavailable data can contribute additional problems in diagnosing and mitigating system faults. Transients may, however, develop over a short period and
may not grant the operators enough time to perform a diagnosis and take the appropriate corrective actions to avoid more serious plant conditions. Also, the amount of knowledge required to be proficient in a particular diagnostic domain, the high levels of diagnostic accural.:Y and speed required, and the limited availability of qualified experts has led to the necessity for the development of artificial intelligent systems designed to aid the diagnostician in his task. The two artificial intelligence (AI) methodologies which have consistently proven themselves to be useful in the development of diagnostic systems, are expert systems, and neural networks.