SHAP-based insights for aerospace PHM: Temporal feature importance, dependencies, robustness, and interaction analysis

This research addresses a critical challenge in aerospace engineering: enhancing the interpretability of machine learning models for predictive maintenance. By integrating SHapley Additive exPlanations (SHAP), our approach decodes the relative importance of sensor-derived features, providing an analytical foundation for understanding engine degradation signals. We delve into the temporal shifts in feature relevance, unveiling the variable impact of sensor data over operational cycles. A rigorous assessment of SHAP's robustness further strengthens the reliability of our interpretive models in the face of data perturbations. Additionally, our nuanced analysis of feature interplay offers a comprehensive view of the factors influencing engine performance predictions. These methodological advancements equip engineers with precise, actionable insights for preemptive maintenance scheduling, directly contributing to the enhancement of aircraft safety and efficiency. The study's implications extend beyond theoretical analysis, offering a pragmatic blueprint for the application of SHAP in the ongoing pursuit of model transparency and maintenance optimization in the aerospace sector. •Innovative SHAP Metrics: Introduced novel evaluation metric Mean Absolute SHAP Values (MASV) to dynamically assess feature importance across different operational phases in aircraft engine lifecycles.•Temporal Feature Importance: Demonstrated that the importance of specific features, such as ‘Nc’ and ‘Ps30’, varies throughout an aircraft engine's lifecycle, providing insights for predictive maintenance.•Temporal Feature Importance Engine-specific: Introduced a novel approach to analyze SHAP values across different operational cycles for individual engines, offering a granular view of temporal feature dynamics in aerospace engine degradation.•Interpreting SHAP Dependency Plots for Predictive Maintenance in Aerospace: Elucidated the practical implications of SHAP dependency plots in predictive maintenance, providing actionable insights for maintenance engineers in the aerospace industry.•SHAP Resilience Analysis: Proposed a framework to quantify the robustness of SHAP values, revealing a correlation between features with elevated SHAP values and high Mean Absolute Deviation (MAD) and Maximum Difference (MaxDiff) metrics.•Model Comparison: Comparative analysis of XGBRegressor and MLPRegressor models revealed nuanced differences in feature importance perception and sensitivity to noise duri