Estimation of Pressure Leakage Severity for Space Habitats Using Extended Kalman Filter

Extraterrestrial habitat systems face numerous disruptions throughout their lifecycle, which may threaten crew safety. One serious risk is associated with air pressure leakage, which can occur due to sudden events such as micrometeorite strikes or airlock failures. To ensure proper functionality of the habitat, it is essential to detect the leak and assess its severity, quantified in terms of the leakage area, and communicate that information to the health management system in a meaningful way to support decision-making. A quantitative prediction of the remaining time before the detected fault results in system failure; its time-to-critical is also helpful for making decisions in such a resource-constrained environment. This paper presents a systematic study to develop and examine methods to estimate the current state of a pressurized vessel and predict the time-to-critical. To achieve this estimate in a real-time setting, an augmented extended Kalman filter method is used in conjunction with two noise identification procedures, namely, expectation maximization and variational Bayes. The robustness of the extended Kalman filter method, whether used independently or in conjunction with the expectation maximization or variational Bayesian approaches, is evaluated through a numerical study. The numerical study yields practical strategies for tuning the algorithm hyperparameters so as to obtain more rapid, accurate, and unbiased estimates of the fault intensity and the remaining time-to-critical. The performance and efficiency of these strategies are then experimentally validated by determining the leakage area and time-to-critical in a lab-scale pressure vessel.