The role of AI in maintenance and asset management
"Industrial AI can revolutionize maintenance by turning reactive processes into proactive and predictive strategies. By using digitalization as a tool, we can make decisions based on facts rather than gut feelings," notes Ramin Karim.
AI Factory Project - a platform for industrial AI
One of the most exciting initiatives in this area is the AI Factory Project, run by Ramin's research group at Luleå University of Technology. The project combines research in asset management and operations and maintenance engineering with a technical platform for developing and testing AI algorithms.
"The AI Factory Project is a package of several research initiatives where we work on developing algorithms and analyzing how AI can be used in different industrial contexts. Our projects are often conducted in collaboration with industry and funded through various research institutes as well as by Vinnova, Sweden's innovation agency that supports research and development in the business sector, Formas, a government research funding agency focused on sustainable development, and the Swedish Research Council, which is responsible for funding and promoting Swedish research of the highest quality," Karim explains.The three key issues in AI and industrial maintenance
To understand the role of AI in industrial maintenance and asset management, organizations should ask themselves three basic questions:
- Are we doing things right? - This is about efficiency and optimization of existing processes.
- Are we doing the right things? - A more strategic question that requires a holistic approach and the right priorities.
- How do we distinguish between right and wrong? - At the highest management level, it's about bringing in the business perspective and making decisions that benefit the business in the long term.
"These questions are relevant to all industries and AI can help us answer them with fact-based analysis instead of subjective estimates," says Karim.
Model sharing - a solution for secure data management
One of the major challenges in AI development is data management. Many organizations are hesitant to share their data for business or security reasons. A new concept in AI development, developed in one of Ramin's projects, is model sharing instead of data sharing.
"Instead of organizations sharing their raw data, they can execute an AI model locally and only send back the result. This reduces the risk of data being used in the wrong context and at the same time provides higher data protection," Karim explains.
Nowcasting and forecasting - a new analytical framework
More and more AI frameworks are being developed to understand the present and plan for the future. One of the most important is the concept of nowcasting and forecasting:
- Nowcasting focuses on analyzing the current health and performance of the plant, rather than just failure rates and historical data.
- Forecasting is about predicting future health in order to take action before problems occur.
"You can compare it to healthcare and health care. Nowcasting is about diagnosing existing problems, while forecasting is about preventing future problems through prediction," Karim explains.
AI regulation and ethics - a key issue for the future
With rapid technological developments come regulatory and ethical challenges. The new EU AI regulation, which entered into force in December 2023, aims to provide a clearer framework for the safe and responsible use of AI.
"The new regulation is a natural extension of the GDPR and focuses on regulating the AI models themselves rather than just the data management. It's about assessing risks and classifying AI systems before they are implemented," said Karim.
Chat-Con - sector-specific AI for industry
Another interesting development in industrial AI is Chat-Con, a specialized variant of generative AI models that can be customized for specific industries.
"Chat-Con is based on the same technology as ChatGPT, but is specialized for different industry segments such as construction, process industry and aerospace. In addition to handling languages, the model has cybersecurity and bias reduction features, making it safer for industrial use," says Karim.
The future of AI and maintenance
AI has the potential to radically change how maintenance and asset management work. With smart algorithms, secure data management methods and clear regulatory frameworks, AI can help increase efficiency and lower costs for industries globally.
"We see tremendous developments in AI, but for it to be fully exploited we need to balance technology and regulation. We need to speed up regulatory alignment so that we can implement new technologies responsibly," concluded Ramin Karim.
But the big question remains: how quickly can industry adapt to the new technology and take full advantage of AI's potential?
