Data Anomaly Detection in Power System State Estimation Based on Support Vector Machine

To address the significant impact of anomalous data on the performance of power system state estimation, this study proposes a Forecasting-Aided state estimation (FASE) anomaly detection method for power systems based on support vector machines (SVM). Initially, to tackle the issue of poor prediction accuracy in traditional FASE, the method utilizes SVM, leveraging predictive data, measurement data, and estimated values to detect various types of data anomalies, including bad data, load spikes, and single-phase earth faults. Additionally, considering the influence of penalty factors and kernel function parameters on classification accuracy, a grey wolf optimization is employed to optimize SVM parameters, enhancing the accuracy of data anomaly detection while balancing computational speed. Finally, simulation tests on the IEEE 33-bus system demonstrate that the proposed method offers strong universality and real-time capabilities, effectively detecting various types of data anomalies in power systems and improving the performance of state estimation.