Research On Steady-State Optimization Of Industry Process Based On Improved PSO Algorithm

Modern industry is characterized by large-scale and automatic production, in which each parameter is steady in normal process. However, the whole production process would deviate the optimal point because of slow disturbance and aging equipment. To search and maitain the optimum operating condition for the production, steady-state optimization is crucial. At present, the key of the steady-state optimization control is how to build the model and optimize the system parameters by modern optimization methods. Therefore, this paper takes the steady-state industrial production process as a background and researches on PSO(particle swarm optimization algorithm) and its application in steady-state optimization.PSO algorithm is succinct, easy to realize, needing less parameters, and not needing the information of gradient. It has become the new focus in the intelligent field. When facing the problem of the complex multi-peak value optimization control, this algorithm can lead the solution of problem to the local extremum easily. To avoid the PSO trapping into local optimization and accelerate the rate of convergence, this paper upgrades the PSO algorithm through integrating the Tabu Search and Simulated Annealing algorithm based on PSO algorithm, in which adopt convergence factor and index decreasing strategy of inertia weight. The validity is verified by the simulation of the standard testing function.For modern industry system, the control target is multiple, the number of its variance is increasing, and exists many constrains. Therefore, it is difficult to construct strict system model through traditional modeling methods. Radial Basis Function Neural Network(RBF), approaching optimal very well, simple structure, fast study speed, is a typical local neural network. In this paper, the RBF is utilized to estimate the system model, and is verified by nonlinear example. Finally, the process of CHP(Cumene Hydroperoxide) breaking up cell would be taken as the scenario. RBF is used to construct system mode and improved PSO algorithm to optimize parameters without rebuild or extend the original equipments. The simulation results proved that the proposed method can reduce energy consumption, cut cost and increase yield.