Operational Optimization Of Catalytic Cracking Reaction-regeneration System

Catalytic cracking(FCC),as the important secondary processing technology in refineries,mainly produces gasoline,diesel oil and liquefied gas.Reaction-regeneration system is the core unit in the process of catalytic cracking.On the premise of determining the type of feed oil and catalyst,the setting values of the operation variables of the reaction-regeneration system are different,the yield of each product is different.Most of the existing operation optimization schemes of the reaction-regeneration system are designed for a single product oil.They do not consider the product distribution structure according to the market demand,which will affect the economic benefits of the refinery.Secondly,the mechanism of catalytic cracking reaction is very complex,and it is difficult to describe the relationship between each product yield and operation variables with accurate mechanism model.However,the existing data model can’t be updated online,the convergence speed is slow,and it is easy to fall into local optimum.Additionally,the existing operation optimization algorithms of catalytic cracking reactionregeneration system have not been designed according to the characteristics of actual production process,and it is difficult to meet the requirements of operation optimization of catalytic cracking reaction-regeneration system in terms of optimization quality and efficiency.Therefore,it is of great significance to establish an accurate product yield model,design an effective operation optimization algorithm for the operation optimization of the reactionregeneration system under the product distribution satisfying the market demand.This article supported by the National Natural Science Foundation project "The theory and implementation technology of global collaborative optimization operation in refinery production process(61590922)",the research on operation optimization of catalytic cracking reaction-regeneration system.The main work is as follows:(1)Description of operation optimization of FCC reaction-regeneration system.Combining with the process flow of catalytic cracking reaction regeneration system,the decision-making process of operation optimization of reaction regeneration system is introduced,thus the operation optimization problem and its corresponding constraints and operation variables are clarified.Then the mathematical description of the operation optimization model is given.Finally,the emphasis and difficulty of the above-mentioned operation optimization problems are analyzed.(2)An improved ensemble stochastic weight online learning algorithm JADE-ERNN is proposed.Based on this method,the yield models of gasoline,diesel and LPG are established respectively.JADE-ERNN algorithm takes stochastic weighted neural network as individual network,uses negative correlation method to integrate individual networks,and introduces regularization term into the integrated network to prevent over-fitting of the algorithm.Additionally,in order to simplify the process of determining model parameters,JADE,an adaptive differential evolution algorithm with optional external archive,is proposed to optimize the scale of integration,penalty coefficient,regularization parameter,number of hidden layer nodes of individual network,and the optimum value of parameters in the activation excitation function in the integrated stochastic weight network.Finally,the online learning mechanism is used to update the model to improve the accuracy of the model.Based on the improved ensemble stochastic weight online learning algorithm mentioned above,the yield models of gasoline,diesel and liquefied gas are established respectively.The simulation experiments are carried out to compare the JADE-ERNN algorithm proposed in this paper with the existing ensemble learning algorithm ERNN and the modeling methods of product oil yield model GABP and RBF.The experimental results show that the yield model established by the proposed method has better accuracy.(3)Considering the characteristics of operation optimization in this paper,an improved differential evolution algorithm named iDE is proposed.Firstly,considering there are many constraints in the FCC operation optimization,and the feasible ranges of different operating variables varies greatly,the cooperative interaction mutation strategy is designed to generate mutants,improving exploitation and exploration capabilities of the algorithm.Secondly,a parameter adaptation strategy with self-repairing capability is proposed to update crossover probability and scaling factor.Additonally,considering that the quick requirement of the operation optimization in FCC unit,an enhanced search strategy is presented to further exploit the best individual in each generation,enhancing the convergence speed of the iDE algorithm.Based on the above-mentioned product yield models,10 different FCC reaction-regeneration system operation optimization test cases were designed.The simulaiton experiments on solving 10 test cases are carried out to compare iDE algorithm proposed in this paper with some representative difference algorithms and FCC operation optimization algorithms.The experimental results show that iDE algorithm has stronger global optimization ability,robustness and faster convergence speed.