Uncertainty-Quantified Regression-Rate Prediction for Hybrid Rocket Motors via Bayesian Fusion

To reduce the design costs of large hybrid rocket motors and enhance the accuracy of regression-rate predictions, this paper presents a regression-rate prediction method based on Bayesian multisource data-fusion theory. By constructing a multisource sample database that incorporates scaled-down motor test data, scaled-down motor simulation data, and large-motor simulation data, the regression rate for large motors is predicted and verified experimentally. Through the uncertainty analysis method, this study systematically investigated the uncertainty factors affecting both simulation and experimental results and quantitatively determined the corresponding uncertainty ranges of the outcomes. Then, the Bayesian data-fusion methodology was further extended to accommodate uncertainty-containing datasets, leading to the establishment of a hierarchical combustion rate fusion framework through which two-stage data fusions were systematically performed. The first round integrates the scaled-down motor test data with the simulation data. The results from this fusion are used in conjunction with scale effects to preliminarily predict the regression rate of the large motor. The second round of fusion integrates the predicted regression rate of the large motor obtained through scale effects with the simulation data of the large motor. This process resulted in the final predicted regression rate and the associated prediction error for the large motor. The prediction results show an error of 1.73% compared with the motor test data, thus indicating higher accuracy than both the simulation calculations and scale-effect extrapolation. Additionally, the model provides the uncertainty range for the predicted regression rates, which contributes to enhancing the reliability and robustness of engine design solutions.