The Research On Controling Method During Mechanical Wentilation Process In Grain Storage

China is the world’s largest grain-producing country, with annual average grain production of0.5or0.6billion tons. With social development and population growing, the demand for food on the market is increasing, too. So it is a challenge for food reserves. Mechanical ventilation is the key technology for grain storage. This paper focuses on study of related control methods to reduce energy consumption and labor costs when making a decision during mechanical ventilation process. So it has a certain research value.The whole study work includes two aspects. Firstly, the heat and mass transfer mathematical model has been studied. Based on the model predictive control method, a PSO-genetic hybrid optimization algorithm was developed. Using rolling optimization method, this proposed hybrid optimization algorithm was used to control grain temperature and humidity to desired value. At the same time, during control process, the overall energy consumption was also optimized. Secondly, SVM technique and its optimization methods had been studied. SVM was used to train machine learning model, which matches up the grain historical data with high accurate rate. When new grain data comes, the model can make the prediction accurately.In the first part of the paper, the importance of grain safe storage was illustrated. Some common methods for food storage were showed briefly. The necessity and implications of mechanical ventilation were illustrated elaborately. At the end of this part, research objectives, tasks and main research contents were introduced. In the second part of this paper, the mechanism of heat and mass transfer model was described in detail, and the discretization of the model was also described elaborately. Instruments and equipment used in the experiment were described briefly. The advantages and disadvantages of particle swarm algorithm and genetic algorithm were discussed. A particle swarm-genetic hybrid optimization algorithm was designed. The detailed process of PSO algorithm was showed. In MATLAB environment, programs based on PSO algorithm and PSO-genetic hybrid optimization algorithm were carried out respectively. In the last part of this paper, the advantages and disadvantages of SVM algorithm were analyzed. Moreover, a comparison between SVM algorithm and expert system was made. Besides, SVM models and relevant optimization algorithms were studied. On the basis of theoretical study, the raw grain data was pre-processed and converted into SVM training format. These processed grain data were used to train SVM model. The built SVM model was proved to be effective and accurate by new testing grain data.All in all, based on grain prediction model, grain storage mechanical ventilation process was optimized with low energy consumption. In addition, SVM technique was applied to the research on massive grain data.