Boreas —
Temporal Convolutional AHU Controller
Boreas predicts optimal supply air temperature setpoints and economizer damper positions for rooftop air handling units. Named for the Greek god of the north wind. It reduces energy waste by anticipating thermal loads 15-30 minutes ahead using zone temperature trends, outdoor conditions, and occupancy patterns.
Abstract
Background
Boreas predicts optimal supply air temperature setpoints and economizer damper positions for rooftop air handling units. Named for the Greek god of the north wind. It reduces energy waste by anticipating thermal loads 15-30 minutes ahead using zone temperature trends, outdoor conditions, and occupancy patterns.
Approach
The model was trained in AxonML (a pure-Rust deep learning framework) and compiled through the Hailo Dataflow Compiler (DFC 3.33.1) targeting Hailo-8 silicon. Post-training INT8 quantization was applied during the DFC compilation pass with production telemetry calibration data. The resulting Hailo Executable Format (HEF) binary executes on Hailo’s fixed-function dataflow architecture with deterministic latency and zero framework overhead at the edge.
Results
On production hardware (Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289)), Boreas achieves 5,391 FPS (hw_only) with 0.224 ms hardware latency at 1.21 W average power draw.
Conclusion
Boreas is production-ready as a single HEF binary deployed to edge devices with no external dependencies beyond the HailoRT vendor runtime. The model meets real-time latency requirements for its target hvac control application.
| Model | Boreas |
| Domain | HVAC Control |
| Architecture | Temporal Convolutional AHU Controller |
| Target silicon | Hailo-8 |
| Measured on | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
| DFC compiler | 3.33.1 |
| Framework | AxonML v0.6 (pure-Rust, CUDA + CPU backends) |
| Author | Andrew Jewell Sr. · ORCID 0009-0005-2158-7060 |
| Organization | AutomataNexus LLC · Fort Wayne, Indiana |
Executive overview
Boreas predicts optimal supply air temperature setpoints and economizer damper positions for rooftop air handling units. Named for the Greek god of the north wind. It reduces energy waste by anticipating thermal loads 15-30 minutes ahead using zone temperature trends, outdoor conditions, and occupancy patterns.
Network I/O
Input: 18-channel AHU sensor array. Output: supply air setpoint + economizer damper position.
Architecture
Temporal Convolutional AHU Controller
6-layer dilated TCN (dilation 1/2/4/8/16/32) with 18 input channels: zone temperatures (multiple), supply/return air temps, mixed air temp, outdoor air temp/humidity, filter differential pressure, fan VFD feedback, damper position feedback, CO2 levels. Hidden dimension 64 throughout. Output predicts supply air temperature setpoint and economizer position.
Compilation constraints
All AxonML models targeting Hailo silicon are compiled under the fixed-function dataflow constraints: no dynamic control flow, no variable-length dimensions, all activations representable in INT8 after calibration, and no operations requiring dedicated softmax hardware (replaced with ReLU gating or depthwise convolution equivalents where necessary).
Silicon performance
Measured on production hardware via hailortcli benchmark with 5-second sustained inference. Device: Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289).
| Metric | Measured Value |
|---|---|
| FPS (hw_only) | 5,391.44 |
| FPS (streaming) | 5,391.26 |
| HW Latency | 0.224000 ms |
| Power (streaming avg) | 1.20800 W |
| Power (streaming max) | 1.20900 W |
| Power (idle) | 0.74979 W |
| Quantization | INT8 (post-training, DFC calibration) |
| DFC Compiler | 3.33.1 |
| HailoRT | 4.20.0 |
| Measured On | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
Deployment
Deployed as a single HEF binary. No ONNX runtime, TensorFlow Lite, or Python inference stack required at the edge.
| Target silicon | Hailo-8 |
| Measured on | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
| DFC compiler | 3.33.1 |
| Quantization | INT8 (post-training, production telemetry calibration) |
| Runtime | HailoRT (vendor runtime) |
| Edge platform | Raspberry Pi 5 + Hailo AI HAT+ (M.2 Key M) |
Deployment procedure
Copy the .hef binary to the target device. hailortcli run loads the HEF directly into the Hailo-8 dataflow engine over PCIe. Inference begins immediately with deterministic per-frame latency. No model conversion, graph optimization, or warmup phase required.
References
- Jewell, A. (2026). AxonML: A Pure-Rust Deep Learning Framework for Edge Inference. AutomataNexus LLC. Technical whitepaper.
- Hailo Technologies Ltd. (2024). Hailo Dataflow Compiler User Guide. DFC v3.33.1.
- Hailo Technologies Ltd. (2024). Hailo-8 Product Datasheet.
AutomataNexus LLC · Fort Wayne, Indiana · andrew.jewellsr@automatanexus.com
Andrew Jewell Sr. · ORCID 0009-0005-2158-7060
May 2026 · All rights reserved.