SRT-VI Type Cracking Furnace Based On The Model Of Control And Optimization

Complex industrial process modeling and optimization is one of thehottest research points in process control fields, and also the hardest work toapply control theory to the industrial process of realities. Cracking furnace isthe key factor of ethylene productions. Its stability, safety and effectivenesswill heavily affect the whole process. Therefore, application of APC(Advanced Process Control) and Operation Optimization to cracking furnacehas great theoretical and practical significances. In order to realize thecracking furnace advanced control, operation optimization task, need onlinemeasurement of cracking furnace yield in the production often through theconfiguration industrial chromatographic analyzer for cracking furnace ofethylene and propylene online yield value. However, industrial chromategraph often have the problem of considerable measurement lag, and theequipment investment is large, maintenance complex, operation cost is high,the failure rate is high. This paper starts according to the complicated processof ethylene cracking, combining with the cracking process, and proceeds withstudies of how to apply PSO (Particle swarm optimization) and Elman NeuralNetwork to model the complex process of pyrolysis.In this paper the selection of a large petrochemical company SRT-VI typeof ethylene cracking furnace control system of the reconstruction project asthe background, the selection of ethylene cracking furnace of cracking depthas the research object. In the ethylene production process modeling, controland optimization technology research status and soft measurement model ofthe existing modeling methods based on the analysis of the deeply, and put forward a kind of adaptive dynamic level particle swarm algorithm(ADHPSO), this algorithm keep particle diversity, can get rid of the localextremum, good global convergence. Then, will ADHPSO applied in ELMANneural network training, the establishment of the ethylene cracking furnacecracking depth online prediction model, this paper puts forward an integratedADHPSO-ELMAN process modeling of cracking depth intelligentoptimization control method, get cracking process optimal operatingconditions. Finally, the simulation calculation by using Matlab, and thesimulation results show that the proposed method is significantly improvedthe yield of ethylene and propylene, has a good stability and adaptability, tothe actual production has great application potential.