Parameter Identification and Situation Awareness in Distribution Networks Under Incomplete Information Driven by Data-Physical Fusion

Some distribution networks are characterized by weak grid structures and complex feeder deployment, which leads to missing information such as line parameters. Existing power flow methods depend heavily on accurate parameters and lack sufficient dynamic adaptability, making them unsuitable for incomplete-information conditions. To address this problem, this paper proposes a distribution-network situation-awareness model driven by data-physical fusion. First, to cope with the dual challenges of incompleteness and uncertainty in line parameters, a bounded multi-operating-condition line-parameter identification model is constructed, and the covariance matrix adaptation evolution strategy is employed to determine line resistances and reactances; the resulting parameters are then used for data augmentation. Next, a situation-awareness model for distribution networks is developed by combining a physics-informed neural network with a mixture-of-experts architecture. In this way, data-driven learning is integrated with physical constraints so that the perceptual capability of deep learning is preserved while the model remains guided by power-flow physics. Finally, the model is validated using the IEEE 33-bus system and historical operating data from an actual distribution area in China. The results show that the proposed model outperforms the comparison methods on the IEEE 33-bus system and effectively tracks multiple state variables in the actual distribution area, supporting its feasibility for incomplete-information conditions.