ABSTRACT The detection of anomalies and prediction of remaining useful life (RUL) of aircraft engines is crucial for ensuring flight safety, optimizing maintenance, and reducing operational costs. Thus, this work proposes a Dual‐Rate Physics‐Guided Stochastic Graph Prognostics Framework for robust anomaly detection and precise RUL prediction in turbofan engines. During anomaly detection, dual‐rate cascading failures behavior hides early precursor signals and leads to a dormant fault mystification effect and reduced reliability. To tackle this, a Stochastic Koopman–Lyapunov Graph Bayesian Network (SKL‐GBN) is presented, which decomposes nonlinear dynamics into dominant temporal modes and captures slow and detects rapid stability shifts and fault correlations, thereby mitigating chrono‐spatial fault blind spots and enabling timely anomaly warnings. Moreover, RUL models try to cluster engines by latent failure modes and rely on global time‐series similarity while ignoring component‐level fault deterioration trajectories, thus creating a granular degradation factor gap. To address this, a Component‐Aware Graph Monte Mixture Network is presented, which independently encodes component‐level degradation trajectories, models inter‐component dependencies and fault propagation paths, and then adapts across distinct degradation regimes. Thus, disentangling severity factors and closing the granular degradation prediction gap in turbofan engine health management. Finally, Multi‐Objective Harris Hawks Optimization (MOHHO) selects the optimal configurations balancing detection accuracy, RUL error, latency, and energy for lightweight edge deployment. As a result, the proposed model attains a high accuracy of 98.3%, an F1‐score of 98.63%, and a low RMSE of 2.18, compared to existing models.
Pani et al. (Sun,) studied this question.