An Iterative Modeling And Trust-region Optimization Method For Batch Processes

Batch process optimization is of great significance in industrial applications. Considering the difficulty of obtaining industrial process models in Model-Based Optimization method and lack of global information without a model in Model-Free Optimization method in traditional methods, this paper proposes an Iterative Modeling and Trust-region Optimization (MTO) method for a typical class of batch processes with characteristics of fast and low cost. IMTO method is a combination of local modeling, trust region optimization and on line experiments. The method works out with high efficiency and low experimental cost. The main contents of the research work are summarized as follows:The key factors of the IMTO method and IMTO algorithm are proposed and illustrated in detail. The key factors contain iterative modeling technique, trust-region optimization and updating and termination criterion. Iterative modeling technique is used to ensure that the local model accuracy is gradually improved. There are three important aspects, model structure selection, data set updating and model residual verification. Trust region optimization is used to ensure the credibility and effectiveness of the iterates. The step length is limited by the trust region in solving the optimization subproblem iteratively.The characteristics of the IMTO method are demonstrated through several numerical simulations of batch processes. Simulations are successfully conducted. Two kinds of typical optimization problems of batch processes are discussed, where surrogate models, replacing the industry batch process, are used as objective functions and constraints, respectively. The detailed analysis and results show IMTO method can work out both of the problems.The IMTO method is also successfully implemented in the quality control and operation optimization of a typical batch process-injection molding process. Similarly, two kinds of typical optimization problems are discussed, where the surrogate model, replacing the industry batch process, is acted as objective function and constraint, respectively. The result shows satisfactory performance and high efficiency. In addition, the method is also compared with the traditional methods under the same operation condition, and the experimental results strongly validate that IMTO method can perform better in solving such quality control and operation optimization of batch processes.