Application Of Improved PSO In Fed-batch Fermentation Process Control

Fed-batch fermentation is the most popular way in fermentation industry, in fed-batch fermentation process, matrix or nutrient are feed to fermentation liquid by intermittent or continuous mode.The control of the feeding rate has an important effect on the yield of fermentation. Traditionally, people usually choose a constant feeding rate according to the experience. However, this constant speed feeding methods are difficult to adapt to the different periods of fermentation process, therefore it is of great significance to obtain a suitable microbial fermentation feeding plan to improve the product yield by selecting a appropriate algorithm to optimize each feeding period.Particle Swarm Optimization(PSO) is a swarm intelligence optimization algorithm. The PSO is simple and robust so that it is suitable for a variety of optimization problems. In traditional PSO, the strong purpose character easily lead to diversity disappears, which gives rise to the particle falling into local extreme points. In addition, the traditional PSO has a poor local searching ability due to the particle can’t iterative search neighborhood. Three improvement strategies were put forward in this paper:In order to improve the diversity of initial population, we used a logistic chaotic sequence to generate the initial particle; Introduce an adaptive speed variation strategy to avoid the disappearance of population diversity caused by stalled particle; Introduce cloning selection mechanism in which we can have a quality search nearby the superior particle in PSO algorithm to improve the local search algorithm. We integrated the three strategies and presented an adaptive speed variation immune particle swarm optimization algorithm IV-PSO. In theory, the improved algorithm has better search performance and the capability of escaping from local extreme points.Two kinds of optimization strategies have been presented in fermentation process optimization control. In WBO strategy, I attempted to maximize the yield of each moment periods, on the contrast we ignored fermentation product yield of each moment and just focused on the final yield. In order to prove the effectiveness of the IV-PSO algorithm, this paper gave the optimization testing of HPV protein and Interleukin-2in two optimization control strategies. The results proved that IV-PSO algorithm has got the maximum yield of protein. What’s more the speed of convergence is superior to other algorithms. In WBO strategy, because of the variable dimension is low, the optimization process is relatively easy, and therefore all optimization algorithms had achieved good results of optimization. In BBO strategy, the variable dimension is high and the optimization results of IV-PSO algorithm are superior to other optimization algorithms. Due to the limited optimization ability the results of the traditional PSO algorithm are even worse than the constant speed control strategy. Integrated all the experimental results, the engineering application value of the improved optimization algorithm presented by this paper has been verified.