Temperature Modeling Study Of Storage Grain Transverse Ventilation Process Based On Support Vector Machine

The safe storage of grain is the key work and measure to guarantee the security of grain supply. Stored grain mechanical ventilation is the most widely used form of grain storage operation, has made a significant contribution to safe storage of grain. According to the real demand of intelligent storage ventilation, this paper studies the intelligent forecasting method based on support vector machine(SVM), which has a certain theoretical and practical value to improve mechanical ventilation efficiency and ensure the grain security.In this paper, the relevant research in the field of grain ventilation and drying are summarized, and the main influencing factors in the process of storage grain ventilation are analyzed. By dividing the grain into sections,this paper takes the average temperature data of each layer from lateral ventilation experiment of Qingyuan National grain depot as the modeling object, uses the traditional optimization method to search for the best parameters, establishes a SVM prediction model, and verifies the feasibility of the model. Meanwhile, The gray relational analysis method is introduced to reduce the dimensionality of the data and improve the prediction performance of the model.Aiming at the limitations of traditional parameter optimization methods, genetic algorithm(GA) and particle swarm optimization (PSO)are applied to improve the traditional SVM model. The simulation results show that the optimized regression model can well fit the non-linear relationship between grain temperature and other variables, which is suitable for predicting the strong non-linear process of storage ventilation.In order to enhance the model anti-noise performance, this paper combines the fuzzy clustering with support vector machine theory, uses Fuzzy C-means clustering algorithm (FCM) to complete the data fuzzification, in order to strengthen the ability to resist interference. The improved fuzzy S VM model has better prediction ability than other models.Finally, this paper selects the experimental data from the same grain depot in another ventilation time of the transverse ventilation as a validation data set, compares the performance and verifies the validity of the above models, the results show that the models are valid and the fuzzy support vector machine model which is optimized by particle swarm optimization has the best performance.