Modeling For Clarifying Process Of Carbonation Method Sugar Factory Based On Gra-Lssvr

Carbonation clarifying process has complex mechanism and many variables, which is hard to build the model. Among variables, some are physical chemical variables; some are production indexes of clarifying process. The indexes can reflect clarifying effect and quality of sugar. Predicting indexes of clarifying process improve production effect. Sugar factory immaturely apply the research theory of predicting so that it mainly rely on human experience to operate. Therefore, this article will build the predicting model on production indexes of clarifying process on the basic of support vector machine to afford a new way for the prediction research of clarifying process.This article firstly confirms production indexes and parameters and preprocesses clarifying data to prepare for the future research. Secondly, calculate the correlation degrees between production indexes and parameters on the basic of gray correlation analysis get the main influencing parameters of the indexes as the input of the model. Thirdly, introduce SVM theory to analyze the principle and algorithm of LSSVR and build separately LSSVR models by the way of many choice and particle swarm optimization algorithm to predict sugar color value and sugar calcium salt and prove the influence of model in data preprocess. In addition, based on the model of PSO-LSSVR, optimize the process set value which is compared to analyze the predicting effect from their initial set value. Finally, use VB to build the modeling software of carbonation clarifying process and show the predicting result of the model on it.The simulation result shows that data preprocessing can improve the effect of the model. The predict effect of the model builded in PSO is better than many choice. Besides, the optimized process set value is better than the initial values to afford a new building model way for the carbonation clarifying process.