Research on Polyurethane Creep Prediction Model Based on BWO-BP

Abstract Polyurethane foam materials can creep under the influence of temperature, humidity, and pressure during use, which seriously affects their safety and state of use. Therefore, studying the creep behavior of polyurethane foam materials and developing corresponding creep mechanism models have significant scientific value for understanding their damage patterns. As a type of machine learning, Backpropagation (BP) neural network can be used for data analysis. However, there are many issues when determining the network structure. Whale Optimization (BWO) algorithm, as a type of swarm intelligence algorithm, can efficiently determine the framework of the BP model. BP which has been optimized through the BWO algorithm, has been utilized to construct a model for the prediction of polyurethane’s creep failure. The optimization process has significantly lowered the mean squared error from 0.30 in the non-optimized BP model to 0.07 in the enhanced BWO-BP model, which validates the effectiveness of the proposed algorithm.