ABSTRACT The widespread deployment of Industrial Internet of Things (IIoT) infrastructures in modern power systems has introduced large‐scale, heterogeneous generation‐side data streams with varying quality, semantics, and reliability. Under the emerging Industry 5.0 paradigm, generation scheduling must be not only automated but also human‐centric and resilient, placing strong demands on trustworthy, transparent, and actionable data processing in real‐time decision loops. However, inconsistencies across multi‐source data can propagate through dispatch constraints and lead to erroneous operational actions. To address this challenge, we propose an IIoT‐enabled framework for trustworthy parsing and cross‐source consistency verification of generation‐side scheduling data. The framework integrates semantic parsing, confidence estimation, provenance‐weighted robust fusion, and constraint‐graph‐based inconsistency modeling into a unified online pipeline. An adaptive trust update mechanism dynamically reweights data sources based on streaming residual evidence, enabling rapid isolation of drifting or faulty sources while preserving reliable ones. Experiments on four public datasets—ENTSO‐E, OPSD, PJM‐5 min, and NREL‐WIND—demonstrate consistent improvements in parsing accuracy, fusion quality, and inconsistency localization, with practical streaming latency suitable for real‐world deployment.
Zheng et al. (Mon,) studied this question.