.png?sfvrsn=ff199933_15)
From Logs to Insights: Streamlining Aircraft Maintenance with Cognitive Advisor powered by AI
Collaborator
SIA Engineering Company (SIAEC) is a leading provider of aircraft maintenance, repair, and overhaul (MRO) services in Asia-Pacific, supporting a global clientele of airlines and aerospace companies. Headquartered in Singapore, SIAEC offers comprehensive line and base maintenance solutions, component support, and engineering services, leveraging decades of operational excellence and strategic partnerships with original equipment manufacturers (OEMs).
SIA Engineering Company (SIAEC) is a leading provider of aircraft maintenance, repair, and overhaul (MRO) services in Asia-Pacific, supporting a global clientele of airlines and aerospace companies. Headquartered in Singapore, SIAEC offers comprehensive line and base maintenance solutions, component support, and engineering services, leveraging decades of operational excellence and strategic partnerships with original equipment manufacturers (OEMs).
Challenges
Commercial aircraft are subject to stringent maintenance regulations to ensure flight and passengers’ safety. Among these is a policy mandating the resolution of recurring faults and defects. Failure to address these can eventually ground the aircraft. Identifying recurring faults involves combining vast volumes of technical logs containing free-text descriptions of faults and the maintenance actions taken. With hundreds of aircraft operating thousands of flights daily, this manual process is time consuming and error-prone. Detecting long-term trends is also difficult due to the ad-hoc nature of faults, variations in writing styles, and the rotation of flight and maintenance crews.
Commercial aircraft are subject to stringent maintenance regulations to ensure flight and passengers’ safety. Among these is a policy mandating the resolution of recurring faults and defects. Failure to address these can eventually ground the aircraft. Identifying recurring faults involves combining vast volumes of technical logs containing free-text descriptions of faults and the maintenance actions taken. With hundreds of aircraft operating thousands of flights daily, this manual process is time consuming and error-prone. Detecting long-term trends is also difficult due to the ad-hoc nature of faults, variations in writing styles, and the rotation of flight and maintenance crews.
Solution
To overcome this challenge, SIAEC’s maintenance and system engineers collaborated with researchers from A*STAR Institute of High Performance Computing (A*STAR IHPC) to develop Cognitive Advisor — an AI-powered solution using embedding-based language models. These models offer a robust and scalable way to analyse large volumes of unstructured text with minimal pre-processing.
A*STAR IHPC led the development of Cognitive Advisor, which included fine-tuning a language model on historical technical logs and building an intuitive front-end GUI (graphical user interface) for seamless interactions with the results. This enabled engineers to cluster recurring defects more effectively, improving troubleshooting efficiency with 30% time savings, and potentially avoiding delays or reducing their duration. Concurrently, SIAEC supported the development by contributing domain expertise and real-world data, which helped guide model training and validate outputs in an operational context.
The solution is adaptable and continues to improve as more data becomes available. Unlike traditional systems reliant on manually curated dictionaries, this data-driven model learns patterns and relationships directly from the text. Its performance is further enhanced when paired with domain knowledge from maintenance experts — offering a superior alternative to existing commercial tools, which are heavily dependent on manually pre-processed keyword-based approaches and are limited in accurately identifying recurring defects.
A*STAR IHPC led the development of Cognitive Advisor, which included fine-tuning a language model on historical technical logs and building an intuitive front-end GUI (graphical user interface) for seamless interactions with the results. This enabled engineers to cluster recurring defects more effectively, improving troubleshooting efficiency with 30% time savings, and potentially avoiding delays or reducing their duration. Concurrently, SIAEC supported the development by contributing domain expertise and real-world data, which helped guide model training and validate outputs in an operational context.
The solution is adaptable and continues to improve as more data becomes available. Unlike traditional systems reliant on manually curated dictionaries, this data-driven model learns patterns and relationships directly from the text. Its performance is further enhanced when paired with domain knowledge from maintenance experts — offering a superior alternative to existing commercial tools, which are heavily dependent on manually pre-processed keyword-based approaches and are limited in accurately identifying recurring defects.
Overview of Cognitive Advisor solution.
Key Results
The initial development focused on a single family of narrowbody commercial aircraft leading to internal deployment for use by SIAEC engineers in 2022. Encouraged by the promising results, the solution was subsequently scaled to cover additional three families – spanning both narrowbody and widebody models. With positive feedback received, SIAEC’s engineers have plans to incorporate Cognitive Advisor to support day-to-day engineering operations.
The initial development focused on a single family of narrowbody commercial aircraft leading to internal deployment for use by SIAEC engineers in 2022. Encouraged by the promising results, the solution was subsequently scaled to cover additional three families – spanning both narrowbody and widebody models. With positive feedback received, SIAEC’s engineers have plans to incorporate Cognitive Advisor to support day-to-day engineering operations.
Significant Progress
This project represents a significant milestone in SIA group’s broader digitalisation efforts. By automating a traditional manual and labour-intensive process, the solution enhances operational efficiency and contributes to the continued competitiveness of the organisation in a fast-evolving aerospace industry.
This project represents a significant milestone in SIA group’s broader digitalisation efforts. By automating a traditional manual and labour-intensive process, the solution enhances operational efficiency and contributes to the continued competitiveness of the organisation in a fast-evolving aerospace industry.
A*STAR celebrates International Women's Day
From groundbreaking discoveries to cutting-edge research, our researchers are empowering the next generation of female science, technology, engineering and mathematics (STEM) leaders.
Get inspired by our #WomeninSTEM