Evolvable Artificial Intelligence for Predictive Maintenance

Predictive maintenance can reduce the effort and cost of maintaining industrial production systems. By using AI to analyze sensor data, equipment can be used more efficiently, running until its predicted end-of-life, avoiding premature replacements. This improves both the economic and environmental sustainability of production systems.

However, the most accurate AI methods require labeled data, such as past equipment failures or health trends, which is often scarce because machines are designed to avoid failure.

This project aims to create an adaptable AI framework for predictive maintenance that continuously improves by using data from similar machines, expert-labeled anomalies, and synthetic data. The platform will be tested in the forest/paper industry.