Optimization of Process Parameters for Induction Welding of Composite Materials Based on NSGA-II and BP Neural Network

In this paper, a process parameter optimization method of induction heating temperature field was proposed, in which finite element simulation as data supply, back-propagation (BP) neural network as data prediction, and non-dominated sorting genetic algorithm II (NSGA-II) as data optimization. Firstly, the electromagnetic-thermal coupling analysis in COMSOL was used to calculate the temperature field of the welding interface of electromagnetic induction welding of CF/PEEK composites, and the temperature data of each point of the welding interface was obtained. Then, the BP neural network model based on the L-M algorithm was established to predict the temperature field of the welding interface, and train and test the established neural network model, got a prediction model that reasonably reveals the welding process parameters and the temperature field relationship of the welding interface. Then, the multi-objective optimization algorithm NSGA-II was used to optimize the design of welding process parameters, and according to the thermal properties of polyetheretherketone (PEEK) resin, based on the crowding calculation and combined with the priority solution analysis method, the priority solution non-dominated sorting genetic algorithm (pNSGA-II)was proposed. Finally, a set of process parameters was obtained that satisfies the optimal heating effect of the weld interface.