<p>The data types of power customer network terminals are diverse, including logs, traffic, and alarms. However, the quality of data from different sources is uneven, and the labeling workload is large, making it difficult to ensure the accuracy and integrity of all data. Therefore, an intrusion target identification method based on a knowledge graph and LightGBM algorithm is designed. The power customer network data is collected, the target frame size of the dataset is analyzed by the K-means clustering algorithm, the feature screening of the data set is performed by the IFSFOA algorithm, the knowledge graph technology is introduced into the intrusion target mining, and the knowledge graph is combined with the LightGBM algorithm. The intrusion target information base constructed by knowledge graph is used as the training data of the LightGBM algorithm to accurately identity the security intrusion target of the power customer network terminal. The experimental results show that the output frequency of the design method is higher than 51.00 Hz, the waveform is smooth, and the alarm rate is highest at 0.99. When the number of nodes is 180, the packet loss rate reaches its the highest value (0.12%), and the throughput is higher than 220 Mbps.</p>