The International Conference for Condition-Based Maintenance in Aerospace (ICCBMA)

The International Conference for Condition-Based Maintenance in Aerospace (ICCBMA), held in Paris from September 11–13, 2024, serves as a prestigious forum dedicated to advancing Condition-Based Maintenance (CBM) methodologies within the aerospace sector. This conference brought together leading researchers, industry experts, and academics to share innovative developments aimed at improving the reliability, safety, cost-efficiency, and environmental impact of aerospace operations. Key themes included structural health monitoring, prognostics, and fleet management, with a particular focus on bridging the gap between theoretical advancements and their practical application within aerospace maintenance.

At ICCBMA 2024, Fatemeh Hosseinpour, a postdoctoral research fellow at Amsterdam University of Applied Sciences (AUAS), in collaboration with Asteris Apostolidis of KLM Royal Dutch Airlines and Konstantinos P. Stamoulis, Professor at AUAS, presented a study titled “Deep Learning-Based Explanations for Remaining Useful Life Prediction of Aircraft Turbofan Engines.” This research project addresses the critical challenge of accurately predicting the Remaining Useful Life (RUL) of aircraft engines, an essential component of predictive maintenance strategies aimed at preventing costly and potentially disruptive engine failures. The study employed advanced Deep Neural Network (DNN) architectures, specifically Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, to model the complex, non-linear degradation trajectories characteristic of turbofan engine performance data. To enhance model interpretability, Explainable Artificial Intelligence (XAI) techniques, notably SHAP and LIME, were integrated. These techniques provide insights into the relative influence of individual variables on RUL predictions, making the complex models more transparent and suitable for high-stakes aviation contexts.

The presentation received significant recognition from conference participants, who emphasised the methodological rigour and practical relevance of the research for aerospace maintenance. Through the effective combination of DNNs with XAI approaches, Hosseinpour and her colleagues not only achieved superior predictive accuracy over traditional models but also delivered valuable insights into the model’s decision-making processes—an essential factor for the adoption of advanced predictive techniques in safety-critical applications. This research marks a significant step forward in the field of predictive maintenance, offering the potential for more reliable, cost-effective, and operationally efficient maintenance frameworks across the aerospace industry.