Open Access Open Access  Restricted Access Subscription Access

Edge-AI for Solar Energy Forecasting: Accuracy–Efficiency Trade-offs using Quantization and Knowledge Distillation

Rohit R Bhat, Dr. Manish Kumar

Abstract


Accurate short-term solar power forecasting is critical for grid management, but state-of-the-art models (e.g., deep LSTMs or ensembles) are too large for on-site microcontrollers. We propose an accuracy-efficiency trade-off: training a lightweight model (small RNN/CNN) via knowledge distillation and post-training quantization, achieving comparable accuracy to a heavy baseline while slashing computational load by 90%. Using a public solar PV dataset [1], our heavy TensorFlow model attains high accuracy (RMSE ≈ 215 kW, R² ≈ 0.949) [2]. The distilled/quantized model’s RMSE remains within 5–10% of the teacher’s, while inference latency and memory footprint drop by 80–90%. This demonstrates edge-deployable solar forecasting with minimal accuracy sacrifice.


Full Text:

PDF

References


Kaggle (2024) Solar Energy Power Generation Dataset. Available: https://www.kaggle.com/datasets/stucom/solar-energy-power-

generation-dataset.

M. Naghapushanam, B. Jeevarathinam and C. Sankari on Physicsinformed voting ensemble for solar power generation forecasting: integrating domain knowledge with machine learning, vol. 9.Energy Inform 9, 3 (2026). https://doi.org/10.1186/s42162-025-00604-7.

B. A. Villamizar-Medina1, A. J. Soto-Vergel2, B. M. D. O. CardozoSarmiento4, D. Guevara-Ibarra5, O. Lo´pez-Bustamante Universidad F. de Paula Santander, Cu´cuta Colombia University at Buffalo, Buffalo, United States, “Neural network quantification for solar radiation prediction: An approach for low power devices,Aibi research, management and engineering journal“. Volume 13, Number 1, Pag 11-19, january 01,

I. R. Nijhum , T. Akinola, Md E. Ali, S. Dev, “OPTIMIZING LIGHTWEIGHT PHOTOVOLTAICFORECASTINGMODELS:A DISTILLATION APPROACH”.

Geoffrey Hinton, Oriol Vinyals and Jeff Dean, “Distilling the Knowledge in a Neural Network,” Neural Information Processing Systems (NIPS) Deep Learning Workshop, 2015.Available: https://arxiv.org/abs/1503.02531

Xingjian Shi, Zhourong Chen, Hao Wang, Dit-Yan Yeung, Wai-kin Wong and Wang-Chun Woo, “Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting,”Advances in Neural Information Processing Systems (NeurIPS), 2015.

Ian Goodfellow, Yoshua Bengio and Aaron Courville,“Deep Learning,”

MIT Press, 2016. Available: https://www.deeplearningbook.org

Andrew G. Howard et al., “MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications,”arXiv preprint arXiv:1704.04861, 2017.

Benoit Jacob, Skirmantas Kligys, Bo Chen et al.,“Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018.

Jian Zhang et al., “Deep Learning Based Solar Power Forecasting: A Review,” Renewable and Sustainable Energy Reviews, 2021.


Refbacks

  • There are currently no refbacks.