<p>The paper addresses the most severe issue of False Data Injection Attacks (FDIAs) with the possibility of cascading failure in unbalanced distribution networks. The attacks have the potential to cause severe threats to the power network reliability and stability. The paper suggests a new detection scheme based on the K-Nearest Neighbors (KNN) algorithm to solve the issue. Unlike conventional approaches that utilize either real-time or historical information alone, the proposed approach creates a feature vector from both the data to identify key system parameters such as voltage levels, phase angles, real power, and reactive power between buses and phases. The features are normalized and labeled according to functional states to enable the KNN model to learn prevalent behavior and attack schemes. The detection of attack is carried out using steady monitoring and distance-based classification, and the detection of the attack is determined by KNN-based majority voting. Metrics like accuracy, precision, recall, and F1-score are utilized to verify that the method performs better than other methods in FDIA detection. It improves the stability and robustness of power grid networks, primarily if they happen to be unbalanced. This high-density solution considerably minimizes the risk of cascading failures and cyber-attacks, guaranteeing permanent operation of interdependent power systems.</p>