Key Parameters Optimization Of Edible Oil Refining Process Based On Data Driven

The refining process of edible oil is an essential process in the production of edible oil.In the refining process,phospholipids and free fatty acids in the crude oil were removed by saponification reaction,and the product oil meeting certain quality standards was obtained.In order to ensure the oleic acid price,soap content,phospholipid content and other indicators of the finished product are qualified,and improve the oil yield as much as possible,the key process parameters such as caustic soda content,phosphoric acid content,centrifuge opening and light phase pressure need to be adjusted in the refining process to reach the production set value suitable for this batch of crude oil.In the production process,when the test value feedback oil is unqualified,it needs to be adjusted according to the observed feedback state.The quality of the adjustment process is directly related to the economic and technical indicators of the whole refinery.The most direct and effective method to improve the refining process is to optimize the parameters.Aiming at the problem that it is difficult to realize the optimization of production process parameters by manual adjustment,this paper proposes an optimization method of refining process operation parameters based on big data:(1)The historical data splicing table in the form of {oil quality +process parameters---quality index + oil yield} is established,and the upper and lower approximation set method of rough set is used to mine expert regulation rules;the constraint range of each parameter corresponding to the highest oil yield in history is mined as the constraint range of parameter optimization.(2)Taking oil yield and product quality as output,crude oil quality and process parameters combination as input data,xgboost model is trained,virtual field is established,and current oil yield is predicted in real time based on online production data.(3)The optimization model of operation parameters is established,and the xgboost prediction model is taken as the virtual field(optimization target).In the parameter optimization interval,the genetic algorithm is used to optimize the process parameters combination of xgboost model with the highest oil yield under certain crude oil quality.The expert rules are used to evaluate the optimization results,and the results that violate the rules can not be distributed as decision-making,so they can be re optimized.According to the field data verification results show that the proposed method can give more accurate parameter adjustment decisions,and can improve the oil yield when the crude oil supply is stable.