Research And Application Of Reactor Vessel Temperature Control Prediction Algorithms Based On Particle Swarm Optimization Algorithm

Chemical industry is the pillar industry of the national economy. As one of the importantchemical product, Ethyl Acetate is commonly and widely used in organic chemicals andindustrial solvents. The EA (Ethyl Acetate) esterification process consists of four primaryprocedure: the reaction vessel stage for esterification, the kettle stage for neutralization, thesieve tower stage for extraction of EA, and the catalytic distillation packing tower stage forextractant recovery. Esterification is happened in the reaction vessel, in which EA is produced.During the esterification process, the quality and the amount of EA are directly affected by theesterification temperature in the vessel. Therefore, the reactor vessel is chosen as the maincontrol object, and a control system is designed to optimize the temperature of the reactionprocess, which is significant to guide practical production, and to improve the quality and theamount of EA.Due to the complexity of the mechanism of esterification reaction process, the presence ofendothermic and exothermic reactions, and the properties of large inertia, nonlinear, modelstructure uncertain and other shortcomings, so the traditional model-based control algorithm isoften ineffective. Thus, on the basis of domestic and foreign study of batch reactor temperaturecontrol of the basic chemical production process, the characteristics of the controlled object areanalysed, then a hybrid control strategy is proposed, which combines predictive controlalgorithm and fuzzy control algorithm, and then the particle swarm optimization algorithm isapplied to optimize the selection of the control rules of fuzzy controller. Following simulationanalysis and application results, the good control performance of this algorithm are proved.Key points of main work are as follows:1, The composition structure of EA production process and the reactor are described; Thereactor temperature control characteristics are analysed; then a system model based onfirst-order lag pure links is established. Respectively the least square method, genetic algorithmand particle swarm algorithm for the model parameters identification are adopted, and theidentification results are compared to determine the PSO (Particle Swarm Optimization)algorithm based on the recognition results as model parameters. According to the lowsystem-dependent and nonlinear mathematical model advantage of the predictive control and fuzzy control, a hybrid control strategy which combined predictive control and fuzzy control isproposed.2, According to the process control requirements, a predict-fuzzy controller while using PSO(Particle Swarm Optimization) is designed on the basis of the temperature rising characteristicsof the reaction process. The PSO algorithm is applied to optimize the selection of the controlrules of fuzzy controller, and to the implementation of the reactor temperature control, avoidingthe traditional learning process of fuzzy control rules from explosions and other shortcomings.The final theoretical simulation verifies that the control method has high reliability, smallovershoot and fast response time advantages.3, The reactor vessel on the EA production line in our laboratory is chosen as one controlledobject, and the Siemens S7-300PLC controller as the slave computer, the Kingview (6.53)software for the PC monitoring platform. Ultimately, the control cabinet system designing,hardware designing and debugging wiring are accomplished. Intelligent control softwarepackage based on STEP7is designed to display dynamic process, data reporting, real-time andhistorical trends such as human-computer interaction screen recording, and to achieveintelligent control reactor temperature. Actual operating results show that the control methodhas high reliability, small overshoot and fast response time characteristics.