Research And Design For Fault Diagnosis Expert System Based On Artificial Neural Network

In recent years, the intelligent systems composed by methods such as probabilistic reasoning, fuzzy logic, neural networks and evolutionary computation has been used widely in Medical care. electricity, finance and other fields. Especially, the Expert System composed of probabilistic reasoning and neural network plays a significant role in practical work of many industries and has made a great achievements. For example, the Fault Diagnosis Expert System used in Power Field, the Disease Diagnosis expert system used in Medical field, the Investment Forecasting Expert System used in financial field and so on.Expert system had such a wide range of applications, because the expert system has the advantages of their own decisions, the expert system relies mainly on the two logical reasoning and decision tree techniques to simulate human reasoning model. Therefore, the expert system has benefits, being good at the knowledge expressed, tolerance of uncertainty data and logical explanation. However, the expert system is still exposed to its inevitable shortcomings, including being bad at tolerance of uncertainty data, data processing, self-learning ability, data mining and system maintenance.The reason why expert systems have shortcomings of one kind or another, tracking the source, is because of the database design patterns of expert system. In designing of expert system knowledge base, knowledge is summarized to rules that are input to the rule base one by one. When that is completed, the expert system itself did not the ability of changing the existing knowledge and adding the new knowledge, unless a Knowledge Engineer is willing to do these works. If you want to solve the shortcomings of expert systems, you will change the rules pattern of knowledge base with the pattern of artificial neural network.Artificial neural network is a parallel data processing to simulate the behavior of the human brain, it has a powerful self-learning ability and is easy to Maintenance and has a great ability of tolerance and inclusion and is good at processing to incomplete data. More important is the artificial neural network has data mining capabilities and the ability of knowledge acquisition, that expert system don't have, these ability are just make up expert system inadequacies and break the "bottleneck" of expert system. Now. implanted the artificial neural network technology to expert system, in order to upgrading the traditional expert system to artificial neural network expert system, which it combines two technologies of artificial neural and expert system,so that it has a strong functions. If a suitable Training algorithm is used in the training of artificial neural network, the expert system will make a great of progress in its performance and become a Self-learning expert system. This expert system will be more "Wisdom", its problem-solving is more closer to human's.In Chapter 1, expert systems will be introduced in a brief case; in Chapter 2, the conventional expert system will be designed; in Chapter 3, artificial neural network expert system will be designed; in Chapter 4, tow different kinds of expert systems will be compared; in Chapter 5, the summarizing about the design of expert system will be done and the outlook of expert system will be putted forward.