Applying Multi-Fidelity Optimization for Process Parameter Optimization in Polymerization Process of Carbon Fiber Production

Carbon fiber is an innovative strategic material in key fields such as national defense, etc. In recent years, the demand for carbon fiber in the entire international and domestic markets is in a period of rapid development. The polymerization process is one of the crucial processes in carbon fiber production. In order to improve the quality of precursor fiber and save costs, the optimization of process parameters is one of the important links. Considering the complexity of process parameter optimization, we improved an optimization method based on multi-fidelity model to solve the problem, which combined the advantages of multi-objective mechanism model and data-driven model of carbon fiber production polymerization process. On the basis of modified ordinal transformation and optimal sampling framework, we embedded a heuristic algorithm to search solution space and combined the clustering algorithm for grouping. The experimental results show that the improved multi-fidelity optimization method outperforms other multi-fidelity optimization methods. This paper can provide guidance for the optimization of process parameters of carbon fiber polymerization, and has a certain reference value.