Research On Blending Scheduling And Operation Optimization In Petrochemical Enterprises

Petroleum refining and metal smelting are the typical representative production process in the process industry.The material blending scheduling,which is at the both ends of the refining process,is directly related to the continuity of production process,product quality and enterprise net profits;Operation optimization can help enterprise to reduce energy and material consumption in the metal smelting process.The research on blending scheduling and operation optimization in petrochemical enterprises has important theoretical and applied significance.In thesis,the main work includes two aspects.On the one hand,the limitations of petrochemical production scheduling problem is described by using mixed integer nonlinear programming methods,and the crude oil and refined oil blending scheduling problem is solved by using generalized disjunctive programming(GDP)through mathematical modeling and optimization.On the other hand,for the complexity mechanism of germanium smelting process,data analysis is used to solve the problem about operation optimization in process of germanium smelting.Details are as follows:1)The continuous-time GDP model about crude oil mixed scheduling problem is established and solved by conversion.The intermittency of the arrival of crude oil and the continuity of crude distillation unit's production is considered in the model.The numerical experiments show that GDP model can greatly shorten the problem solving time.2)The discrete-time and continuous-time GDP model about the refined oil blending scheduling problem is respectively established and solved by conversion.In the continuous-time model the relationship of nonlinear component is taken into consideration.The numerical experiments show the effectiveness of GDP model.3)A least squares support vector machine(LSSVM)algorithm is presented,which is based on estimation of distribution.The regularization parameter and kernel parameter of LSSVM model,are optimized by estimation of distribution algorithm.Results of prediction show that the algorithm can meet the demand of practical production?4)One decision support system for germanium metal balance and prediction is developed,which is based on the production background of one germanium smelting company.This system can realize electronic management of metering data of production process,analyze the reasons for the loss of material to find out bottlenecks in production,and predict metal grade of germanium dust by calling LSSVM algorithm embed in the system.GDP model of refinery blending scheduling can be applied to the actual refining enterprises for the actual production management decision.At the same time,GDP method can be extended to the relevant scheduling problem.The decision support system for germanium metal balance and prediction can be directly applied to germanium production process achieving optimal control of production.