Study On Operation Optimization Of Petrochemical Process Based On Improved Genetic Algorithm

Optimize operating parameters in real time based on process simulation data to realize operation optimization of petrochemical production process is an important method for improving product profit,saving energy and reducing consumption in the current petrochemical production process.Among them,Genetic Algorithm(GA)as an optimization calculation tool plays an important role,but the randomness of its mutation operation is too large to achieve optimal direction search,and produce a lot of invalid operations,resulting in low GA operation efficiency and other problems.To solve this problem,this thesis embeds the Pseudo Partial Derivative(PPD)concept of Model-Free Adaptive Control(MFAC)theory into GA,utilize PPD gradient to guide the mutation operation of GA,and restrict individual fitness with expectations in the meanwhile,so it can achieve orientation mutation and individual fitness optimization.The main research includes:(1)Utilizing the MFAC method to calculate the optimal PPD value of each individual and the mutation value of the next individual,and then get the fitness of each individual of this generation.Then make the individual mutate in the direction of the minimum difference between the individual's fitness and the expectation by the guidance of PPD,so-called the direction of PPD gradient.So it can realize the orientation mutation.(2)Set individual fitness expectations to restrict the changes of individual fitness.Calculate the individual mutation value after orientation mutation based on the adaptive adjustment characteristics of MFAC and the direction of PPD gradient,and then get the optimal individual fitness.By the guidance of PPD,the fitness of each individual is closer to the expectation,so it can achieve the purpose of optimizing individual fitness.(3)In the research on operation optimization based on simulated data,the variable feasible region is hidden and cannot be expressed explicitly,resulting in a large number of invalid simulations and the simulation cannot converge.In this thesis,the atmospheric distillation column of crude oil distillation unit of petroleum processing is selected as the research object,and focused on the analysis of the impact of system product quality,equipment specifications and tray hydraulics on the feasible region,in order to improving the research on operation optimization of petrochemical production process operation utilizing improved GA based on PPD.Finally,the above improved GA based on PPD is applied to the atmospheric distillation column of a crude oil distillation unit of 2.3 million tons/a in a refinery,take the reflux flow of three pump-arounds as optimization variables and take the total output exergy of reflux flow as objective function.Under the environment of Intel Core i7-10510U 1.80GHz,12.0GB RAM,the calculation shows that the improved GA can reach convergence in the second generation,the program running time is 677 seconds,the maximum total output exergy is505.8×10~4kcal/h.Compared with primal GA,the optimize generation is reduced by 17generations as well as the running time is reduced by 7538 seconds.Under the premise of meeting the constraints of tray hydraulics and product quality,the optimize working conditions will increase the benefit by 8.299 million CNY/a.