MotorLens AI
Electrical telemetry ML modeling for predictive maintenance decisions.
MotorLens AI detects pump and motor problems using electrical telemetry, with a first use-case of cryogenic fluid pump motors. Working with an 89 GB Royal Netherlands Navy dataset containing 348.6 million sample-time rows, we built a decision-tree-based ML pipeline that assesses motor health from just three seconds of electrical data. Our model achieved 90.7% accuracy on unseen recordings from the training motor and 76.0% on a separate motor with different construction and fault conditions, without retraining. These results demonstrate generalization beyond the data used for training. An interactive dashboard explains the findings, demonstrating strong support for future applications in industries such as maritime, rocketry, industrial systems (e.g. data-center cooling).
