Neucleosoft — Electronic Design House, Noida, India
Machine learning built on data from hardware we instrument ourselves
AI and machine learning for electronic products from Neucleosoft: test rigs, data collection and in-house models, including EV battery life prediction.
How we work
Instrument — We build the test bench and instrument the product itself, so the data comes from the real device.
Collect — Bench runs and field runs feed one dataset, instead of two sets of numbers that never meet.
Model — A machine-learning model is trained on that dataset to answer the one question the product needs answered.
Apply — The prediction goes back into the product: what the user is shown and how the device behaves.
Case study: Warivo E Mobility
Predicting how much life is left in a lithium pack (active programme). The bench and the instrumented vehicle feed one dataset. A machine-learning model reads it to predict how much life is left in the lithium pack, so the range a rider sees stays honest as the pack ages.
ML-based RUL prediction — A model estimates the remaining useful life of the pack from the test data.
Range a rider can plan around — The range shown to the rider tracks the pack as it ages.
Optimised charging extends pack life — Charging is optimised to extend the life of the pack.
Built for the programme: Instrumented test vehicle, Test display fitted to the cowl, Battery lab test equipment.