Study On The Optimization Of Polymerization Based On The Co-evolutionary Particle Swarm

As an important branch of chemical control area, polymerization has always been the research hotspot of some domestic and overseas experts. There is a strong exothermic characteristic in polymerization reactors process, it has the characteristic of nonlinearity, time-varying and time-delay. It belongs to difficult control system. The changes of the temperature and concentration in continuous reaction's tank have great influence with the product quality. To ensure the final product quality, every link of the whole process in the production must need strict temperature control.Optimization algorithm provides a good method.With intelligent optimization algorithm development, it provided new ideas and methods for optimization theory, has applied in the industrial, defense and other fields. Particle swarm optimization algorithm is a randomized algorithm based on global search strategy. The concept is simple and the implement is easy, the parameters needs adjustment is small and without gradient information, optimization showed great potential. The theoretical research work is still in its early stages, for complex multi-peak optimal control problem as well as high-dimensional problem, the algorithm easy to fall into local minimum.In this paper, analyzed the basic principles and the advantages and disadvantages of PSO and used the strategies of Co-evolutionary to improve the algorithm, the improved algorithm widely collected the optimal solution of the population, and the features of the optimal solution has a wide range. Comparing to the traditional swarm optimization algorithm, the solution space of the improved algorithm has better global coverage ability, which reduces the probability that the particles into local extremism probability and improves the searching efficiency of the algorithm. By using the same standard test functions to verify, compared with the basic elementary particle swarm optimization, genetic algorithm optimization results. The results show that the improved algorithm is better in the speed and precision than others. Improved particle swarm has been used optimizing refrigerant flow through a continuous stirred tank settlement between the two optimal steady-state transitions.