Guarded Agent-in-the-Loop Residual Adaptation for Evolving Wind Power Regimes

Yuhang Dong; Min Xia; Jinghong Li; Yue Dai; Xinyue Wang; Shi Liang

cs.MAOct 26, 2026

Guarded Agent-in-the-Loop Residual Adaptation for Evolving Wind Power Regimes

Authors: Yuhang Dong, Min Xia, Jinghong Li, Yue Dai, Xinyue Wang, Shi Liang

Organizations: Nanjing University of Information Science and Technology, Nanjing, China

Abstract

Wind power forecasting models are often deployed as fixed temporal backbones, but the errors they make can vary across meteorological regimes such as cold waves, heat-and-calm periods, productive typhoon, and destructive storms. We present a guarded Agent-in-the-loop residual adaptation framework that leaves the temporal backbone unchanged and adapts only a small set of scene-conditioned residual experts. An Agent layer interprets external scene information, identifies the current regime, selects residual experts, and applies a correction only when recent observed errors support it. New experts enter the pool after they outperform their initializations on a validation set, keeping model growth under explicit control. Backbone-only and Agent-enhanced forecasts are compared at lead times of 4, 8, 12, 24, and 48 across nine time-series backbones. Across these runs, the guarded framework lowers both MAE and RMSE for all 45 backbone-horizon combinations, with average MAE reductions ranging from 11.21% to 20.26% across backbones. Online-selection baselines, component ablations, and routing analyses identify the roles of residual experts, scene-aware routing, and the online acceptance rule. Candidate-scene runs evaluate the promotion path under controlled conditions; long-term unattended deployment was not evaluated.

Figures & tables

Figure 1: Guarded Agent-in-the-loop residual-adaptation framework: the backbone models temporal dynamics, while Agents organize scene evidence, route residual experts, guard corrections, and audit candidate-scene updates.

Backbone	Lead	Base MAE	Agents MAE	MAE red.	Base RMSE	Agents RMSE	RMSE red.
PatchTST	4	$202.20\pm 26.44$	$140.13\pm 4.44$	$29.89\pm 7.88\%$	$264.06\pm 31.83$	$206.34\pm 9.60$	$21.15\pm 7.83\%$
	8	$231.83\pm 14.78$	$184.90\pm 3.65$	$19.95\pm 5.93\%$	$301.61\pm 15.21$	$262.93\pm 10.08$	$12.68\pm 4.75\%$
	12	$249.87\pm 8.54$	$213.14\pm 3.89$	$14.61\pm 3.63\%$	$325.47\pm 5.73$	$296.00\pm 3.72$	$9.02\pm 2.37\%$
	24	$258.48\pm 5.56$	$233.10\pm 3.92$	$9.81\pm 0.89\%$	$339.44\pm 2.43$	$316.46\pm 1.81$	$6.77\pm 0.72\%$
	48	$270.68\pm 5.20$	$233.88\pm 5.34$	$13.58\pm 2.13\%$	$367.24\pm 11.60$	$327.02\pm 5.24$	$10.91\pm 2.11\%$
PatchTST-Zoo	4	$183.62\pm 8.40$	$140.29\pm 1.65$	$23.44\pm 4.14\%$	$237.56\pm 5.12$	$204.50\pm 3.55$	$13.87\pm 2.94\%$

Table 1: Backbone-only and Agent-in-the-loop forecasts at all evaluated leads. Values are five-seed mean

\pm

std from the saved paired prediction protocol; reductions are computed per seed against the corresponding backbone-only control at the same lead and then averaged.

Figure 2: Scene-wise error trends for known and candidate wind-power regimes. Orange panels mark known scenes and blue panels mark candidate-scene streams. The $y$ -axis is absolute forecast error, so no observed-power curve is drawn. DFreqLinear is shown as the reference backbone trace, Agents as the guarded residual path, grey points denote other backbone-only controls, and the grey ribbon summarizes their error range.

Figure 3: Multi-horizon Agent-in-the-loop gains at leads 4, 8, 12, 24, and 48. Panel (a) encodes paired MAE reduction by colour and circle area, with black rims marking the best lead for each backbone. Panel (b) compares average MAE and RMSE reductions, and panel (c) shows the online residual acceptance rate used by the guarded correction path.

Variant	Pool	MoE	Guard	MAE red.	RMSE red.
Backbone only	–	–	–	0.00%	0.00%
Residual pool + static expert	Y	static	–	5.28%	3.27%
Residual pool + uniform experts	Y	uniform	–	5.33%	3.25%
Residual pool + soft adaptive MoE	Y	soft	–	5.44%	3.03%
Residual pool + hybrid MoE	Y	hybrid	–	5.50%	3.33%
Residual pool + MoE + online guard	Y	hybrid/learned	Y	15.57%	10.59%

Table 2: Component ablation for the Agent-in-the-loop residual path. The rows separate residual experts, mixture-of-experts routing, and the online guard, showing which part of the Agent layer contributes to the reported gains.

Selector	Coverage	MAE red.	RMSE red.	Max reject run
Backbone only	0.00	0.00%	0.00%	1800.0
Raw Agent	1.00	0.65%	0.55	0.0
Residual ensemble	0.50	2.51%	1.45%	1800.0
Random matched	0.46	0.96%	0.78	11.8
Hedge	0.61	3.24%	0.86%	507.4
DMA	0.45	6.72%	4.35%	330.9

Table 3: Direct online-selection baselines on the same saved residual forecasts. Coverage is the fraction of windows where the residual path is used; DMA coverage is its mean residual weight. Max reject run is the longest consecutive run of forecast windows in which the selector rejects the residual path and falls back to the backbone, averaged over backbone–seed–lead cells. The fixed residual ensemble is not a binary accept/reject selector, so its reject-run value is not interpreted. Oracle uses current-window labels and is reported only as an upper bound.

Figure 4: Main-text selector comparison. Circles show MAE reduction, diamonds show RMSE reduction, and the aligned right panel shows residual acceptance coverage under the same saved paired forecasts. The full coverage-risk and ledger diagnostics are in Appendix Figure 5.

Wind Power Forecasting

Wind power forecasting predicts turbine and farm output from operational data, weather forecasts, and remote observations to support reliable energy planning.

See all 24 papers

UniWind: Day-Ahead Wind Power Forecasting via Physics-Informed State Routing
Jul 2, 2026Ronghui Xu, Tongxin Wu, Guozhen Zhang +4Time Series Forecasting Physics-Informed ML
East China Normal University, Shanghai, China · Southern University of Science and Technology, Shenzhen, China · TsingRoc.ai, Beijing, China +1
Robust Residual Correction via Selective Deployment for Time Series Forecasting
Dec 27, 2025Jianxiang Xie, Yuncheng Hua, Mingyue Cheng +2Robust Time Series Forecasting Residual Learning
School of Computer Science and Engineering, University of New South Wales, Sydney, Australia · State Key Laboratory of Cognitive Intelligence, School of Computer Science and Technology, University of Science and Technology of China, Hefei, China · The Hong Kong University of Science and Technology (Guangzhou)
ECTO: Exogenous-Conditioned Temporal Operator for Ultra-Short-Term Wind Power Forecasting
May 12, 2026Cao Yuan, Junjun WangTime Series Forecasting Non-Stationary Time Series Forecasting
Wuhan Polytechnic University, Wuhan, 430048, Hubei, China

Jul 2, 2026J/K move · Enter open · S save

UniWind: Day-Ahead Wind Power Forecasting via Physics-Informed State Routing

Ronghui Xu, Tongxin Wu, Guozhen Zhang, Yihan Li+3

East China Normal University, Shanghai, China · Southern University of Science and Technology, Shenzhen, China · TsingRoc.ai, Beijing, China+1

Time Series Forecasting Physics-Informed ML Wind Power Forecasting

Open paper AskPDF arXiv

Accurate day-ahead wind power forecasting is essential for cost-effective power-system operation. Wind power generation is primarily driven by future meteorological conditions while retaining temporal dependencies in power generation. In practice, observed wind-farm power reflects both meteorology-driven generation potential and deviations associated with local environmental and operational factors. Existing physical models provide useful constraints but adapt poorly across wind farms, whereas data-driven models can capture rich correlations but do not explicitly structure these two sources of variation. To bridge this gap, we propose UniWind, a wind power forecasting framework that integrates physics-informed generation priors with data-driven state-aware corrections. UniWind first employs a Physical Prior Estimator to construct a site-calibrated physical prior by combining a site-conditioned monotone warp with a shared physical power curve. It further applies a physical upper-bound constraint to shape this prior as a soft envelope of available wind power generation. UniWind then uses a Latent State Encoder to construct latent power-state representations from historical prior-power discrepancies and future meteorological context, and transforms the physical prior into final power forecasts through a State-aware Power Corrector with pseudo-label-guided routing and bounded expert correction. Experiments on 24 real-world datasets demonstrate UniWind's strong forecasting accuracy, robustness, and generalization across full-shot, cross-farm zero-shot, and cross-region zero-shot settings.

Abstract

Wind Power Forecasting

Explore similar work

UniWind: Day-Ahead Wind Power Forecasting via Physics-Informed State Routing

Robust Residual Correction via Selective Deployment for Time Series Forecasting

ECTO: Exogenous-Conditioned Temporal Operator for Ultra-Short-Term Wind Power Forecasting