Research on Intelligent Additive Manufacturing of Mullite Whiskers

Abstract Mullite whiskers exhibit exceptional properties such as mechanical durability, oxidative stability, low thermal expansion coefficient, high-temperature tolerance, and excellent thermal shock resistance, making them widely used in aerospace, mechanical manufacturing, and other fields. This study investigates the synergistic effects of thermal processing parameters and molybdenum oxide catalyst concentration on mullite whisker’s evolution. We propose a predictive model for mullite whisker preparation based on integrating Particle Swarm Optimization (PSO) and Backpropagation Neural Network (BPNN). The model employs a BPNN optimized by the PSO algorithm, leveraging historical experimental data to predict the influence of varying temperatures and molybdenum oxide concentrations on whisker growth without additional experiments. This approach significantly reduces experimental time and costs. Comparative analysis with other nonlinear regression models demonstrates that the proposed model exhibits high accuracy and generalization capability in predicting the impact of temperature and molybdenum oxide concentration on mullite whisker growth. Thus, it represents an effective method for predicting the growth of mullite whiskers.