Operation Optimization Of Naphtha Pyrolysis Process Based On Multi-Objective Evolutionary Algorithm

Ethylene and propylene are the most important monomers in the petrochemical industry,and their production is mainly through the pyrolysis production process with naphtha or alkane as the material.Steam cracking furnace is the main equipment in the process of pyrolysis,which is the highest production capacity in petrochemical industry,and also one of the largest energy consumption devices.The production process of naphtha pyrolysis consists of a large number of complex hydrocarbon cracking reaction,which is a complicated process with highly nonlinear and strong coupling.In the pro-cess of thermal cracking,the increase of yield of ethylene will inevitably cause the de-crease of yield of propylene.The operation level of the production process will directly determine the yields of the products and the cost of the production.Therefore,this paper investigated the multi-objective operation optimization problem of the naphtha pyroly-sis production process so as to help the petrochemical enterprises improve the yields of both ethylene and propylene,as well as the economic benefits.The detailed research topics are as follows:(1)To handle the difficulty of establishing a precise mechanism model due to the complex reactions of naphtha pyrolysis,a least square support vector machine(LSSVM)based on the historical data of the production process was proposed and combined with an evolutionary algorithm to build a data-driven model for the yields of ethylene and propylene.Subsequently the multi-objective operation optimization model of the naph-tha pyrolysis was established to maximize the yields of both the ethylene and propylene.(2)To overcome the disadvantage of loss of diversity of traditional multi-objective differential evolution algorithm(MODE),an improved MODE with memory mecha-nism was developed.In this algorithm,the high quality solutions found during evolution were stored and then used to construct the next population through a new crowding dis-tance that considered the distribution of different Pareto fronts.Computational results based on benchmark problems showed that the proposed algorithm was superior to some powerful multi-objective evolutionary algorithms in the literature.(3)Due to the too random evolution directions in canonical MOEA/D,an im-proved MOEA/D with adaptive scaling factor was proposed.In this algorithm,each sub-problem maintained a sub-population and the correlation between each decision variable in the sub-population and the solution of the single optimization problem ob-tained by the polymerization function was used to update the evolution direction and velocity of each solution.Computational results based on benchmark problems showed that the proposed algorithm was superior to the canonical MOEA/D and some other powerful multi-objective evolutionary algorithms in the literature.(4)The proposed improved multi-objective differential evolution algorithm and the improved MOEA/D algorithm were applied to solve the multi-objective operation opti-mization of the naphtha pyrolysis process,and compared with other multi-objective evolutionary algorithms.The results showed that the proposed algorithms were superior to other evolutionary algorithms in the quality of solutions,and the obtained results had promising practical application value.