Research On The Method Of Improved Neural Network Decoupling In Vacuum Annealing Furnace Application

With the advances in modern industry, heat treatment of materials have becomeincreasingly demanding, vacuum annealing technology has significantly improved thethermal properties of materials．However, vacuum annealing system is a multivariable,strong coupling of time-varying systems; Coupling will reduce the control quality ofthe controlled system, which will not run serious. To control this phenomenon, mustcarry out decoupling control of multivariable systems. The traditional method ofdecoupling control is difficult to meet the requirements of such systems, the neuralnetwork decoupling control has obvious advantages, and thus the neural networkdecoupling control method is introduced into a vacuum annealing system is of greatsignificance.In this paper, a vacuum annealing system modeling analysis, two ways toestablish a vacuum annealing system model, and the reliability of the model forsimulation. The model analysis coupling of the three-zone temperature vacuumannealing. In addition, multiple-input multiple-output system for the decouplingcontrols theory into the synthesis and analysis.On this basis, the paper particle swarm (PSO), the genetic algorithm (GA) andtheir hybrid algorithm is discussed and put forward based on improved PSO-GAhybrid optimization algorithm for training neural network decoupling control method,and the use of Matlab The performance of the decoupling method into the simulation.Combined with vacuum annealing system and the method by simulation Method bysimulation of the effects of decoupling and the decoupling, will this decouplingmethod with the PID neural network decoupling control and PSO algorithm PIDneural network decoupling control methods are compared, the results show that theproposed New methods in controlling decoupling have been a good control effect.