Adaptive weighted dynamic time warping clustering with a deep codec model for accurate photovoltaic power prediction
Zheng Xin, Guangyu Zheng, Weiran Chi, Jianing Lu, Lili Hao · 2026
Read Adaptive weighted dynamic time warping clustering with a deep codec model for accurate photovoltaic power prediction by Zheng Xin, Guangyu Zheng, Weiran Chi, Jianing Lu, Lili Hao (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/PMC13558777.txt.
Description
Zheng Xin and four colleagues forecast in 2026 solar power output minutes to hours ahead by first grouping similar days, then predicting within each group. Ultra-short forecasts keep grids stable as rooftop and farm solar floods in, yet sunlight's flicker defeats simple models. Their clustering step warps time series onto each other with feature weights that adapt, favouring the measurements that vary most, and a pruning step keeps the computation lean.
Each cluster feeds a hybrid predictor: a sharpened temporal convolutional network reads the multivariate patterns and geographic links, while a memory network decodes the longer trend. Tests on real data from a solar plant in Australia beat the benchmark models at every horizon from five minutes to four hours. The authors argue that sorting the weather regimes first is what lets the predictor stay reliable when conditions swing.
Catalog record
| CREATOR | Zheng Xin, Guangyu Zheng, Weiran Chi, Jianing Lu, Lili Hao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 96230 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |