Study On Modeling Of Omethoate Synthesis Process Based On PSO Algorithm

As a great agriculture country,agriculture is the key of national economy of China and has great relation with the country’s stability and the social development. Pesticide is an important factor to ensure agricultural production. The production and quantity of the pesticide have much impact on our agriculture,then on the whole country. To improve the quality of pesticide production,it is very necessary to improve the traditional production process of the pesticide. In this paper, the temperature of Omethoate synthesis process is the research object. Building a better model for the synthesis process will help improve product quality and production efficiency,create the obvious economic benefit.The technique flow of the Omethoate composition is analyzed, and the characteristics of the Omethoate synthesis process and the object model is analyzed. It is clarified that Omethoate synthesis process has the following characteristics:non-linear,time-delay,time-changing and complex disturbance. The influencing factors during synthesis process are analyzed from the measured and unmeasured parameters. Omethoate synthesis is a typical batch process and it cannot be modeled satisfactory. However,neural networks,fuzzy logic and intelligent evolutionary algorithms provided a new approach for modeling such complicated objects.Particle swarm optimization and support vector regression are united to model the temperature object of Omethoate synthesis. Standard PSO algorithm has some shortcomings. The whole population is divided into two parts and each part evolves according to the triangle function inertia strategy respectively. Its performance is tested with benchmark function, the experimental result shows it has faster convergence speed and better searching precision. The machine learning and statistical learning theory are introduced to lead to the principle and method of support vector machine. The cores of support vector machine : the optimal hyperplane and kernel function is introduced and the mathematical model of support vector regression is introduced emphatically. Characteristics of the synthetic Omethoate are analyzed. Combining the improved PSO algorithm and SVR identifies the temperature object. The PSO-SVR model is compared with the static BP net model. It turned out that the model combines the advantages of PSO with SVR. It better reflects actual system dynamic performance. The model has high accuracy and good generalization capacity.