A Study Of Intelligent Optimization Technology And Its Application In Chemical Industry

Optimization can bring great benefits to factories.So it is a focus of study and application nowadays.When referring to chemical industry optimization,it needs to solve huge scale optimization,on-line optimization and multi-objectives optimization more and more on one hand,it has high non-linear characteristic,multi-constrains,dynamic model,random and high safe need which make its optimization difficult on the other hand.In this paper,it studied and developed some intelligent technology in order to meet these needs and try to solve these difficulties. This work focused on solving chemical industry modeling problem and optimization problem,aiming to avoid sophisticated mathematical calculation,accelerate rate of optimizing and improve optimization effect. It also wanted to provide a new point of view of optimization.This work studied mainly on the following aspects and each aspect was studied by the usual sequence of solving optimization problem: making objectives,modeling,designing algorithm,calculation and implementing control.1.It studied chemical industry modeling problem by way of optimization method.Virtual Supervisor-Artificial Immune Algorithm was developed to help modeling batch reactor with nonlinear relationship and unmeasured states based on Structure Approaching Hybrid Neural Networks:using virtual supervisors to train neural networks,optimizing virtual supervisors and exerting benefits of artificial immunity to explore and keep feasible.When modeling process of producing accelerant for sulfuring rubber with unmeasured states,it got a better result.2.It developed Non-dominated Sorting Particle Swarm Multi-objectives Optimization Algorithm with methods of non-dominated sorting,crowding distance,elitism strategy and swarm intelligence to solve multi-objectives optimization problems.Then it was put into optimizing polymerization stage of PET continuous producing aimed to get maximum yield and best quality.A set of Pareto optimal solutions was get and actual producing may be instructed with these solutions.3.It developed Graded Approach for Multi-objectives Optimization algorithm to optimizing batch reaction which is a dynamic process.By using some prior knowledge functions and iterative optimizing prior knowledge functions and parameters of functions,an optimal control trajectory could be approached better.In optimizing simulation process of styrene suspending polymerization,this method reduced number of deciding variables,improved convergent rate and got better multi-objectives optimization solutions.4.Hierarchical intelligent optimization control,iterative optimization control and on-line predict optimization control were reviewed and modified to keep intelligent technology putting into practice.All works in this paper proves that intelligent technology has better effect in optimization and have actual value in chemical industry modeling and optimization control.