Neural-network-based Intelligent Predictive System And Its Application

Artifical Nueral Network(ANN) has been catching the world's attention since it 鑑me into being in the forties of the twentieth century. Now it has been developed to be a useful non-linear processing tool, and is used with great success in many fields. Prediction is one of important application fields of ANN. Since most of the general predicting methods are based on linear analysis, when it comes to non-linear cases they met many difficulties, while ANN is competent for non-linear proceeding for its excellent non-linear character. Predicting methods based on ANN extend the space of predicting research.The following are what this paper has done in the field of prediction especially unemployment prediction.This paper solved some problems in designing the Neural-Network-Based Intelligent Predictive System including data pretreatment, primary factors analysis, and ANN's structure design.Before using ANN to predict, the data have to be pretreated. In some cases, one variable of sample is great different to others in scale, so they have to be standardized in the same scale. And when the 鑑歟 is that samples do not distribute in the same interval, some samples need to be inserted into "blank point". The pretreatment provides the foundation for the follovving prediction.When prediction with little training sample set and large variable si瀍 is concerned, this paper abstracts the prime factors from the training sample set, then only inputs the prime factors into ANN instead of primary variables. This diminishes the si瀍 of ANN with little information loss, which may result in improving the generalization in the 鑑歟 of insufficient training samples.Genetic algorithm is illustrated in the paper to optimize the structure of ANN.And then, the Neural-Network-Based Intelligent Predictive System is designed and used in the research of Chinese unemplovment predicting.According to Chinese unemployment prediction, three models is built in the paper based on Hierarchical Diagonal Neural Netvvork, Diagonal Elman Neural Netvvork, and ANN predictive model following prime factors abstracting.In the end of the paper, the softvvare of the Neural-Network-Based Intelligent Predictive System is compiled with Matlab5.3 version, and used to predict Chinese unemployment. The softvvare achieves good prediction performance.