Study On The Optimization Method Of Process Parameters In Sugarcane Juice Clarification

The clarification process of sugarcane juice is an important section in sugar factories.The quality of the clarification effect affects the production benefit and product quality.At present,sugar factories usually set the process parameters based on manual experience,which is subjective and arbitrary.It is the main reason for the large fluctuations of process indicators in the clarification production process,which seriously affects the production benefit and the quality of finished sugar.In view of the problems that the process indexes of the clarification process cannot be measured online,the lack of a reasonable process index value as the production target of the clarification process,and the lack of basis for the adjustment of process parameters,the following research contents are carried out in this thesis:(1)On the basis of the extreme learning machine,the regression modeling method of the deep kernel extreme learning machine is studied,and the datadriven modeling method of the sugarcane juice clarification process based on the deep kernel extreme learning machine is proposed.The online prediction of two process indexes: the purity difference of the mixed clear juice and clear color value is realized.The effectiveness of the proposed data-driven model is proved by simulation experiments,and the superiority of the proposed data-driven model is verified by comparison with other models.(2)Based on the multi-objective particle swarm optimization algorithm,combining the niche sharing mechanism with the multi-objective particle swarm optimization algorithm,the process indexes optimization method of sugarcane juice clarification process is studied.The optimal compromise solution of the process indexes is obtained,which can be used to guide the sugarcane juice clarification production process and evaluate the quality of the working conditions.(3)The process parameters optimization method based on niche multiobjective particle swarm optimization algorithm and the process parameters optimization method based on case-based reasoning are studied.The optimization speed of the case-based reasoning method is fast,but it cannot solve the problem of case retrieval failure.The niche multi-objective particle swarm optimization algorithm does not depend on successful cases and can find the optimal solution,but the iterations number is large and the optimization speed is slow.Combining the advantages of the two optimization methods,the case-based reasoning method and the niche multi-objective particle swarm optimization algorithm are combined to jointly optimize the process parameters.Finally,simulations verify the effectiveness of the optimization method.(4)Using Matlab GUI to integrate the data-driven model,process indexes optimization model,and process parameters optimization model of sugarcane juice clarification process mentioned in this thesis.A process parameters optimization system for sugarcane juice clarification process is developed,and each model algorithms is packaged into the optimization system.The humancomputer interaction interface is designed to realize the optimization setting of the process parameters of the clarification process.