Trustworthiness Enhancement of Distribution-Level Synchrophasor Data via Multisource Identification Framework

The safety and reliability of distribution-level synchrophasor streaming are crucial in bolstering the strength and resilience of the power grid, particularly prior to initiating data-driven applications such as state estimation, situational awareness, and distributed control. However, malicious attackers can spoof or tamper with source information in the distribution-level system without altering the underlying measurement values. To address this threat, this paper proposes a synchrophasor trustworthiness enhancement strategy grounded in a multi-source identification framework. First, Dynamical Ensemble Decomposition (DED) is proposed to extract distinctive signal signatures from synchrophasors while effectively mitigating modal aliasing. Then, multi-source synchrophasor measurements are integrated to overcome limitations related to low resolution and incomplete information. Next, a Multi-source External Attention (MEA) network is developed to enhance feature diversity through a multi-head attention structure. By integrating DED and MEA, the proposed framework enables robust source authentication. Comprehensive experiments on two real-world power systems across multiple measurement nodes demonstrate that the proposed method consistently outperforms state-of-the-art baselines.