The Application Of Artificial Neural Networks In Regional Weather Forecasting System

Expert System is an important research fields in computer applications. Expert Systems can replace human to work in some circumstances, especially in some dangerous situations.Since the introducing of artificial intelligence technology into expert systems, the performance of expert Systems has been greatly improved and the applications of expert systems have become more widely. For traditional Expert Systems, their frameworks are clear and comprehensible on the basis of the knowledge bases, and on the kernel of the inference machine. Their inference machines work on the basis of the realize conditions, deduce a certain conclusion (or a middle conclusion) depending on their special inference strategies, and then, the middle conclusion puts in the bases of realize conditions, so, the next deducing works as the last, the process will be over until the expectation result comes. The traditional Expert Systems are on the basis of the knowledge bases, they are not fit for those that the expert experiences changes quickly, because that a certain expert experiences is turned into a rule is difficulty, and that a rule join into a rule system is not easy also. This paper researches its internal factors, and finds next: firstly, the knowledge bases and the inference machine are separating in the traditional Expert System, secondly, the knowledge bases must guarantee complete and seal, so, it is difficult that a new join in, and machine self study is not easy.Artificial Neural Networks are an experience models, they emulate some functions of biological neural network. Recently, the methods of Artificial Neural Networks have been introduced into expert system, i.e., expert system based on artificial neural networks, this makes the performances of the expert system improved much more. An expert system based on Artificial Neural Network gives some what intelligences and can inferences by computing, thus, such expert systems can overcome some disadvantages in traditional expert systems. This paper uses ANN to study region weather forecasting. A regional weather forecasting system has been established by artificial neural networks. It shows that all the expert experiences into ANN structure, in this structure, all the know ledges and the inference rules are included together. This suits for the brain thoughts better. In this paper, the test on the similar feather data is done, and the results are good.