The Neural Network Expert Systems and Their Application in Nuclear Reactor Fault Diagnosis
Yang Yiping, Tai Juwei(Institute of Automation, Academia Sirica).
Neural network is a new kind of representation model which differs from the current logical model. In the paper, we apply it to RODES -an expert system for diagnosing nuclear reactor fault and propose the idea of neural net heuristics. The parallelism, distribution and capability of learning of neural net heuristics provide a new way for knowledge acquisition, knowledge representation and uncertainty reasoning in expert systems. The practice in nuclear reactor fault diagnosis shows its potential power in the research of artificial intelligence.