Optimization Method for the PVB Resin Process Parameters Based on the Quality Index Prediction Model

This paper presents an optimization method for the parameters of the synthesis process of polyvinyl butyral (PVB) resin based on a quality index prediction model. First, a PVB resin quality index prediction model is established based on a backpropagation (BP) neural network. The model uses 17 input parameters and predicts 2 output parameters (the melt index and viscosity). Moreover, considering the reaction mechanism, 6 key input parameters are analyzed, and their weights are improved in the model. Then, based on the constructed quality index prediction model, an optimization method for the PVB resin process parameters is developed using the nondominated sorting genetic algorithm II (NSGA-II). Finally, experiments are performed to verify the effectiveness of the proposed method. The results show that under a similar melt index (with a deviation of 3.8%), the viscosity increased by 15.4% after optimization, and under a similar viscosity (with a deviation of 3.0%), the melt index increased by 88.6% after optimization.The proposed method can be beneficial for industrial production of PVB resins that can meet more requirements.