Study On Decoupling Control Based On PSO Neural Network And Its Application In The Distillation

Rectifying column is a important separation equipment in petrochemical industry,which is mainly used for the separation process of multi component chemicalproducts. As a typical multivariable coupling system, the control performance willdirectly affect the energy consumption of distillation production process and productquality. Thus, research a effective decoupling control method and develop a highprecision control system for distillation column has received attention in the field ofchemical process control. Based on the analysis of coupling properties of distillationtower, this paper studies the PID decoupling control method based on neural network.The main work is as follows:(1) On the basis of the research outland and inland, this paper introduce the processand control requirements of the distillation tower, analysis the characteristics ofnonlinear, large delay, multi variable and strong coupling of the tower, put forward aoverall system control scheme.(2) According to the serious temperature coupling of tower bottom and top and theinsufficient of traditional decoupling control method, this paper proposes a neuralnetwork PID control method based on chaos particle swarm optimization. Usingchaos particle swarm algorithm to replace the reverse pass algorithm of original PIDneural network, adjusting the weights of PIDNN between each neuron, the algorithmachieved rapid decoupling control effect. The simulation results show that theproposed method in this paper, compared with the original BP algorithm, has moreexcellent dynamic and steady-state performance.(3) Further analysis of the nonlinear, large inertia, strong coupling characteristics ofdistillation tower, this paper proposes a single neuron PID decoupling controlalgorithm, which is based on dynamic RBF neural network. First dynamic RBF neuralnetworks are built to identify coupling system model. Then Jacobian matrixinformation obtained from identification is used to tune the single neuron PIDcontroller parameters online self-tuning, to further accomplish the decoupling controlof the coupling system. The simulation result shows that the new proposed algorithm has more excellent control precision and robustness than the early existed PIDdecoupling control based on the traditional RBF neural networks.(4) Based on the rectification tower, using Siemens s7-300PLC as lower machinecontroller and the Kingview software (6.53) as the PC monitoring platform, design anintelligent decoupling control system for distillation tower. Complete the design anddebugging of control cabinet hardware system, making the PC configuration interface,developing an intelligent control program based on STEP7software. Doing theexperiment of temperature control of distillation column, results show that thedecoupling control method proposed in this paper has good dynamic performance,high control precision and strong robustness, improve the performance of rectifyingcolumn decoupling control system significantly, and has high practical value.