基于Kriging代理模型的PX氧化过程优化

基于流程仿真模型的复杂化工过程优化，往往需要较长的优化时间，效率低下,因此提出了用拉丁超立方采样和Kriging建模法构造流程仿真模型的代理模型，建立基于Kriging代理模型的多目标优化策略。将该策略应用于PX氧化反应过程优化，结果表明：所建立的Kriging代理模型的3个目标的输出精度都控制在1%以内，建模精度高。采用改进的多目标粒子群算法对此Kriging代理模型进行优化，不但能收敛到全局最优解，而且与PX流程仿真模型优化相比，优化时间大大减少，提高了优化效率。因此，在满足精度要求的前提下，Kriging代理模型可以代替PX流程仿真模型来进行优化，并具有较高的优化效率。

PX Oxidation Process Optimization Based on Kriging Surrogate Model

Based on the process simulation model, complex chemical process optimization often takes a long time and has low efficiency. By using latin hypercube sampling and Kriging modeling method, a multi objective optimization strategy based on Kriging model is proposed, then this strategy is applied to PX oxidation reaction process optimization. The simulation results show that the precision of the established Kriging surrogate model for three goals output is less than 1%. Improved multi objective particle algorithm is used to optimize Kriging surrogate model, not only global optimal solution can be obtained, but also running time is saved obviously compared with the process simulation model. Therefore, Kriging model can take place of the PX process simulation model to realize optimization at the condition of meeting the requirement of accuracy, and has high optimization efficiency.