Operational optimization for large-scale manufacturing processes

 

Developed a modulized hierarchical structure, which included correlative structural model-set and local-generated correlative models, proposed an optimization algorithm that was able to solve the optimal operations in real-time for large-scale manufacturing processes. The algorithm was innovated in the areas of hybrid automatic differentiation, space decomposition, and iterative memory enhancement. A data rectification method was proposed based on Bayes network and hybrid system, which could efficiently solve the consistency problem in the manufacture data. The algorithm was widely used in large-scale ethylene production, PTA production, oil refinery industry and steel manufacturing industry, yielded significant economic and social benefits. As the secondary principle investigator, the project was awarded the 2nd class National Scientific Technology Progress Award in 2005.