Applied Research Of Intelligent Optimization Algorithm In Modeling Of Fermentation Process

Fermentation is a complex process, including biological, physical and chemical changes. With the development of society, the industrial fermentation process arouses people's attention more and more, fermentation scale industrial production also expanded. Improves the product quality, reduces energy consumption, lower production costs, has the significance to the nowadays large-scale development of fermentation industry. Therefore, modeling and optimization have become an imperative for the fermentation industry. Fermentation process modeling is a basic problem to fermentation engineering, it is servers for the fermentation process control and optimization, the accurate model is advantageous in obtaining the better control policy and the optimized method.In recent years, some novel optimization algorithm, such as artificial neural networks, genetic algorithms and swarm intelligence algorithm, by simulation or reveal certain natural phenomena or process development, its content related to mathematics, evolutionary biology, artificial intelligence, neural science and quantum statistics, provide a new way of thinking and means to solve the complex engineering problems.This article in view of the microorganism fermentative process's modeling question, take the glutanic acid fermentative process as the object of study, uses the support vector machines return algorithm to establish the soft survey model, carries on the forecast to fermentative process's three important variables.The standard support vector machines learning algorithm question may sums up to the quadratic form question with restrains. This article in view of support vector machines large-scale training in algorithm convergence rate slow, complex degree higher question, proposed that LPSO algorithm to solve the quadratic programming problem, in view of the LPSO algorithm's in precocious restraining question, proposed the cluster analysis grain of subgroup algorithm (CLPSO) the algorithm. The CLPSO algorithm had guaranteed the particle population's multiplicity, enables the particle to carry on the overall situation search effectively. The experiment and the simulation result indicated that uses the CLPSO training the support vector machines soft survey model to be able the accurate output estimated value.