Small sample data-driven multi-objective optimization of MEMS sensor molding process parameters

In this paper, for the molding defects of MEMS sensors, a multi-objective optimization and decision-making method based on small-sample data-driven molding process parameters is proposed on the basis of MoldFlow software. Firstly, using Latin-square experimental design, each test scheme is substituted into a finite element model for numerical simulation, and a multi-objective optimization model of mold temperature, melt temperature, injection time, pressure-holding time, holding pressure with wire sweep index and Warpage deformation amount, is established by using a random forest algorithm based on the simulation results, and the three hyperparameters of the model are optimized using the grid search method to improve the prediction accuracy. Secondly, multi-objective gray wolf algorithm is applied to optimize the multi-objective optimization model and obtain a Pareto Solution set and finally the subjective and objective comprehensive entropy weight-TOPSIS method is used to evaluate and decide the optimal combination of process parameters. The results show that the correlation coefficients R of the two optimization objectives are greater than 0.98, which indicates that the prediction accuracy of RF model is high, and meanwhile, after the optimization of MOGWO algorithm, the above two quality indexes are improved. The feasibility of the multi-objective optimization method for plastic package molding process parameters based on MoldFlow, RF and MOGWO is illustrated.