Application Of The Cloud Genetic RBF Neural Network In The Grain Situation Predication Model

In recent years, because of the country are very concerned about "three agriculture" problem, implemented a series of favorable policies, combined with the farmers’ scientific farming, make food production for years, the state built many large state-owned grain depot, in which the capacity of single grain depot is several times larger than ones that were built in the past,in a pile of grain reserves of mildew, pest problem is more serious. Food as a special and complicated life, grain piles the changing rule of the internal temperature field is complicated, how to establish the forecast model of grain heap internal situation is facing problems.To achieve the goal of accurate prediction of the change trend of temperature in grain depots, this dissertation mainly focuses on grain heap of ecological environment, study grain situation prediction model to do the following:1) Described the structure model and learning algorithm of RBF neural network and the basic theory, characteristics and shortcomings of genetic algorithm. And for the lack of genetic algorithm is proposed a new algorithm based on cloud theory.,2)Using nonlinear intelligent modeling method based on Cloud model genetic algorithm improved RBF neural network by using of simple mechanism of RBF neural network which Can approximate any nonlinear function and other advantages,at last analysis of the algorithm comparison with other algorithms; Cloud genetic RBF neural network algorithm is applied to the grain bulk temperature forecast, and then study a new algorithm which apply to grain temperature prediction model.3)For detecting the temperature, humidity, moisture, worms and other factors in grain bulk, cloud genetic algorithm is applied to optimize the value of distance threshold ε based on using K-means clustering algorithm to determine the number of the RBF neural network radial function center. Then by using the information detected by information on grain condition decided by temperature, humidity, moisture, worms and other factors, combined with belief function and respective advantages of D-S evidence theory and Cloud genetic RBF neural network and after optimal combined algorithm, information on grain condition fusion optimization based on Cloud genetic RBF neural network is proposed. The detailed experimental result shows that this optimization has better robustness effect on detecting information on grain condition.