Strict Recursive Closed-Loop Evaluation of Greenhouse Air Temperature Forecasting with Generated Exogenous Variables and Exogenous-Error Propagation Analysis
Aiguang Zhang, Shuo Zhang, Jingyu Bian, Jianfei Xing, Wentao Li, Hong Gu, Xufeng Wang · 2026
Read Strict Recursive Closed-Loop Evaluation of Greenhouse Air Temperature Forecasting with Generated Exogenous Variables and Exogenous-Error Propagation Analysis by Aiguang Zhang, Shuo Zhang, Jingyu Bian, Jianfei Xing, Wentao Li, Hong Gu, Xufeng Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13468724.txt.
Description
Aiguang Zhang and colleagues set a harder test in 2026 for forecasting greenhouse air temperature. The model must predict hour after hour without ever peeking at real future readings, because a working greenhouse controller never sees them either. Weather inputs were also forecast and fed back recursively, so errors could compound exactly as they would in deployment. A liquid neural network, a continuous-time model, faced three established rivals under identical data conditions.
Across ten repeated runs, the liquid network posted the lowest one-step error, with mean absolute error of 0.4771 degrees Celsius. The study then analyzed how errors from the generated weather variables propagate through the closed loop. Honest evaluation protocols matter in agricultural control. A model that shines with leaked future data can fail silently in a real greenhouse full of crops. That is the trap this protocol was built to expose.
Catalog record
| CREATOR | Aiguang Zhang, Shuo Zhang, Jingyu Bian, Jianfei Xing, Wentao Li, Hong Gu, Xufeng Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80219 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |