False data injection attack (FDIA) is one of the major cyber-attack patterns facing the state estimation of modern power distribution systems. To defend against false data injection attacks targeting power privacy data security, this paper proposes a distributed data-driven FDIA detection model and solution algorithm for three-phase unbalanced power distribution systems. A polynomial pseudo measurement false data attack pattern for state estimation of three-phase unbalanced distribution systems is discovered by combining the principle of bad data detection and considering the node power privacy data and the attacker’s preference. The distributed solution is performed by the variable penalty coefficient improved alternating direction method of multiplier (-ADMM) method, which protects the data privacy while realizing efficient and accurate FDIA detection and tampered data correction. The effectiveness of the proposed distributed detection algorithm is verified by setting different attack strengths in the IEEE 123-node test system and comparing the analysis with the centralized method and the traditional FDIA detection method, respectively.