Research on the Forecasting Model of Sand-Dust Storm Based on the Grid Field

The sand-dust storm data set can be characterized by high field distribution, high dimensionality and huge data volume, which explains why forecasting results of sandstorms are hardly satisfactory. BP neural network provides a tutor style of learning. Meanwhile, GA is a parallel algorithm based on natural selection and genetic rules, so it is often used in global searching and global optimization. In this paper, a sand-storm forecasting model is constructed and implemented using BP neural network together with GA algorithm. The result of the experiment shows that the GA-ANN approach has higher performances in stability, accuracy and the running speed.