An inverse prediction and experimental study for the internal insulator boundary conditions of solid rocket motor

This paper primarily estimates the boundary conditions of the ablative insulation structure within a solid rocket motor and proposes a non-contact measurement method to mitigate the impact of thermal perturbations and slag deposition on measurement results. The thermal boundary conditions in the combustion chamber severely restrict the refined design of insulation structures. Due to the strong nonlinearity of the heat conduction system caused by the harsh thermal environment, reducing the ill-posedness of the nonlinear inverse problem in non-contact boundary condition measurement has always been a challenge. In this study, a model for solving the nonlinear inverse problem is developed using Physics-Informed Neural Networks (PINNs) and automatic differentiation. Compared to regularization methods, the proposed algorithm reduces the condition number to its square root. Numerical simulations and quartz lamp heating tests show that the algorithm remains stable and robust against noise pollution from thermocouples and data acquisition systems. The method is applied to high overload conditions of a solid rocket motor, successfully measuring the boundary conditions inside the combustion chamber. The obtained data is validated and analyzed through comparison with contact method and numerical simulations to ensure its reliability and rationality.