An integrated Bayesian framework for aircraft performance evaluation from segmented flight tests

Evaluating the full-mission reliability defined by terminal performance metrics from segmented flight test data is challenging due to data fragmentation and complex uncertainty propagation. To address these issues, this paper proposes a Bayesian decision-support framework. For single-test analysis, a Variational Bayesian Particle Filter with a Mixture of Inverse Gammas (VBPF-MIG) is developed for adaptive segment estimation of process noise to infer the posterior distribution of the terminal state such as the miss distance. A Hierarchical Bayesian Model (HBM) fuses sparse statistical summaries to infer the population mean and variance while accounting for between-test heterogeneity. Finally, an adaptive Bayesian sequential test policy utilizes the posterior to dynamically determine the optimal sample size for high-confidence targets. Numerical simulations validate the framework components. The VBPF-MIG method maintains the nominal 95% coverage rate while mitigating the prediction interval inflation of static noise models. The HBM recovers heterogeneous population characteristics, overcoming prior biases when the sample size reaches 20. The adaptive sequential policy reduces the required sample size by up to 23.5% compared to static batch strategies. The tactical missile application confirms the capability of the framework to uncover epistemic limitations in system robustness.