Research On Improvement And Application Of RBF In Forecasting Car Market

With the development of economy and the improvement of people's living standards, the consumption of car has become a hotspot people pay attention to. It is the key point that cars enter into the common people's families, according to the development course of car industry. Meanwhile, it makes car industry important instruments in which nation startups economy and stimulates domestic demand. Therefore, it is necessary to forecast the demand car in the future.It is so difficult that the researchers have spent lots of efforts in study tasks. At present, there are 200 forecast methods, 30 of which are widely used and more than 10 of which are used commonly. This data increases constantly with the advance of time. Various correlative factors having effect on the demand of car market are not simple linear relation but complex non-linear relation. The traditional forecast methods face difficulties when they resolve these problems. It makes the forecast results not satisfied. In order to obtain more satisfied result, we constantly attempt to apply new scientific theories and technologies to reasonable forecast in forecasting the demand of car market. Nerve network method is a science, springing up in recent years. With the development of computer science, its application has become more and more wide. However, the nerve network method is still at the elementary stage of exploration and practice, we find out that radial basis network has better forecast characteristic by the doing research on nerve network method.First of all, this paper utilizes the traditional methods to forecast and analyzes the reasons for the errors from the forecast results and to discover the complex relations between various correlative factors and car demand quantum. Then, using the thought of equal dimensionality vectors improves the radial basis network. We find out that RBF network can resolve the problems that traditional methods can not resolve, through comparing RBF network with traditional methods in forecasting the demand results of car market in the several years' future. RBF network utilizes the principle of artificial intelligence and multilayer information transferred to establish the complex relation between after-variable and before-variable and it has obvious advantage, comparing with the traditional methods. Therefore, the forecast results with RBF network are more exact and reliable.