Do RUL explanations hold up? Faithfulness and stability of attributions on C-MAPSS

2026-10-03

AI Summary

This study addresses the limitations of existing explanation methods for deep remaining useful life (RUL) prediction models in terms of faithfulness and stability evaluation. For the first time, it systematically evaluates deletion/insertion faithfulness and cross-seed stability, explicitly distinguishing noise robustness from retraining consistency. Based on CNN, LSTM, and Transformer architectures, this work comparatively analyzes the explainable AI (XAI) performance of attribution algorithms, including Integrated Gradients, Occlusion, and attention mechanisms. The results demonstrate that Occlusion and Integrated Gradients achieve superior faithfulness, rendering them suitable for practical engineering maintenance reports, whereas raw attention should serve solely as a visualization aid. These findings provide a reliable foundation for interpretable predictive maintenance.

Abstract

Deep remaining-useful-life (RUL) models on NASA C-MAPSS are now routine, and so are heatmaps that colour sensors and timesteps. A heatmap that looks mechanical is not the same as an explanation an engineer can act on. We train three standard architectures - a 1D CNN, an LSTM, and a small Transformer encoder - on the official FD001 and FD003 splits with the piecewise RUL cap of 125 cycles and the official PHM08 asymmetric score. We then attach three attribution maps (Integrated Gradients, occlusion, last-layer attention) and evaluate them with the checks the XAI-for-PdM literature still under-reports: deletion/insertion faithfulness, Spearman stability under sensor-scale noise, agreement across training seeds, and cosine consistency inside RUL bins. Prediction error is a prerequisite, not the claim. The headline is which explanation method moves the RUL output when its top cells are removed, and which map survives a 5% input perturbation. Integrated Gradients and occlusion are similarly faithful on the LSTM; Transformer attention is cheap and temporally smooth but weakly faithful. All three maps are almost unchanged under 5% input noise, yet IG/occlusion agree only moderately across two LSTM seeds - stability to sensor jitter is not the same as stability to retraining. A secondary tabular check on the AI4I 2020 failure dataset shows the same deletion pattern for tree importances. We recommend occlusion or IG for any C-MAPSS-style report that will be read by a maintenance engineer, and we treat raw attention weights as a visualisation only.

Problem

• Remaining Useful Life
• Attribution Faithfulness
• Explanation Stability
• Predictive Maintenance
• Explainable AI

Innovation

• Remaining Useful Life (RUL)
• Attribution Methods
• Faithfulness Evaluation
• Stability Analysis
• Explainable AI (XAI)