Wave prediction tool for a wave-energy converter
A PyTorch and Streamlit tool that predicts wave elevation seconds ahead of a wave-energy converter, using measurements from upstream sensors. It supports Airy, Stokes, JONSWAP and Bretschneider wave spectra and directional wave fields, and exports to LabVIEW real-time targets.
- Capability
- Machine learning and data
- Client
- Atargis Energy
- Sector
- Wave energy
- Stack
- PyTorch, Streamlit, space-time POD, LSTM / GRU, LabVIEW export
Challenge, method, outcome
Challenge
A wave-energy converter controller works better if it knows the incoming wave before it arrives. The task is to predict wave height a few seconds ahead from sensors placed upstream.
Method
Space-time POD reduces the sensor data. LSTM and GRU models then predict the wave at several horizons. The tool is delivered as a Streamlit application with export to real-time hardware.
Outcome
Wave Estimation Tool v1.1 (2026), built for the control loop of Atargis Energy’s wave-energy converter.