Estimation of Behind-the-Meter Solar Generation by Integrating Physical with Statistical Models

Rui Yang, Yingchen Zhang, Farzana Kabir, Nanpeng Yu, Weixin Yao

Research output: Contribution to conferencePaperpeer-review

31 Scopus Citations


Accurate estimation of solar photovoltaic (PV) generation is crucial for distribution grid control and optimization. Unfortunately, most of the residential solar PV installations are behind-the-meter. Thus, utilities only have access to the net load readings. This paper presents an unsupervised framework for estimating solar PV generation by disaggregating the net load readings. The proposed framework synergistically combines a physical PV system performance model with a statistical model for load estimation. Specifically, our algorithm iteratively estimates solar PV generation with a physical model and electric load with the Hidden Markov model regression. The proposed algorithm is also capable of estimating the key technical parameters of the solar PV systems. Our proposed method is validated against net load and solar PV generation data gathered from residential customers located in Austin, Texas. The validation results show that our method reduces mean squared error by 44% compared to the state-of-the-art disaggregation algorithm.


Conference2019 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids, SmartGridComm 2019

Bibliographical note

Publisher Copyright:
© 2019 IEEE.

NREL Publication Number

  • NREL/CP-5D00-74647


  • data models
  • estimation
  • hidden Markov models
  • inverters
  • load modeling
  • solar power generation
  • system performance


Dive into the research topics of 'Estimation of Behind-the-Meter Solar Generation by Integrating Physical with Statistical Models'. Together they form a unique fingerprint.

Cite this