Aviation safety remains a primary concern in the era of modern air transportation. On Thursday (12/2), Adryan Fitra Azyus officially earned his doctoral degree from the Faculty of Mathematics and Natural Sciences (FMIPA) at the Universitas Indonesia (UI) with the distinction of very satisfactory, after presenting his dissertation entitled “Prediction of the Remaining Useful Life of Jet Aircraft Engines Using the CNN-GRU (Convolutional Neural Network – Gated Recurrent Unit) Method.” The open defense was held at the Prof. Dr. Siwabessy Auditorium, FMIPA UI, and was chaired by the Dean of FMIPA UI, Prof. Dr. Tito Latif Indra.
This research emerged from major challenges in aviation safety, particularly concerning turbofan jet engines that operate under extreme conditions and are prone to material degradation. Many aircraft accidents originate from engine failures caused by improper maintenance procedures or inspection shortcomings. Although such disasters are relatively rare, their impact places significant pressure on the aviation industry and regulatory authorities to strengthen regulations and oversight.
Physically, engine degradation occurs due to the accumulation of material changes such as fouling, erosion, and creep, which lead to microcracks and gradual alterations in the metal structure over time. Unfortunately, conventional maintenance methods often fail to detect early signs of damage because they assume linear behavior in systems that are inherently non-linear. This poses a significant challenge in accurately predicting the engine’s Remaining Useful Life (RUL).
In his research, Adryan sought to bridge the gap between the principles of failure physics and modern technology by developing an accurate and efficient Deep Learning–based method for predicting Remaining Useful Life (RUL). He integrated the principles of Predictive Maintenance (PdM) with a hybrid CNN-GRU architecture capable of learning hidden patterns in engine sensor data while simultaneously forecasting degradation history. The CNN component extracts inter-sensor correlations to understand spatial relationships, whereas the GRU models temporal degradation, enabling the engine’s performance history to be captured with precision.
“This research is not merely about calculating numbers or predicting machine performance. Its primary goal is to ensure passenger safety and airline operational efficiency by predicting failures before they occur,” said Adryan. This approach is referred to as Spatio-Temporal because it is capable of capturing the physical relationships of the engine across both space and time simultaneously.
This research utilized the C-MAPSS dataset from NASA, which provides run-to-failure data and models interactions among engine components based on principles of energy and mass. The dataset enables the model to learn real degradation patterns during engine operation, unlike other datasets that focus solely on control systems or flight dynamics.
“With this CNN-GRU technology, we can detect engine degradation patterns that are typically hidden from manual inspections, enabling maintenance to be carried out in a timely and more efficient manner,” Adryan added.
The research findings show that the CNN-GRU architecture is capable of handling high noise levels in real operational data—an issue that conventional methods often fail to address—and can adapt to six different operating conditions. The use of a two-gate GRU mechanism makes model training 11.98% faster than the industry-standard LSTM, without compromising accuracy. This enables real-time RUL prediction with minimal computational load.
The research supervisor, Dr. Budhy Kurniawan, emphasized that this innovation provides a tangible contribution to the aviation industry, as it can help airline operators minimize the risk of engine failure and optimize maintenance scheduling. Co-supervisor Drs. Sastra Kusuma Wijaya, Ph.D., added that the findings also support industry sustainability by reducing material waste caused by excessive maintenance. Adryan aims for this technology to be implemented in real-time aircraft fleet management systems, enabling operators to perform maintenance proactively before failures occur.
The open defense was also attended by the Head of the Indonesian Halal Product Assurance Agency (BPJPH), Dr. Haikal Hassan, ST., MT., as well as a member of the Advisory Board of the Sharia Economic Society (MES) and former President Director of Bank Muamalat, DR (HC) Ahmad Riawan Amin. This achievement reflects the commitment of the Faculty of Mathematics and Natural Sciences (FMIPA) at the Universitas Indonesia to advancing research that is relevant to industry and societal challenges, while further strengthening the faculty’s reputation as a center of scientific excellence and technological innovation at both the national and international levels.
