Bad Data Identification in Power System Based on Improved GSA Algorithm

Based on the analysis of the improved GSA (gap statistical algorithm) data mining technology, it is applied to the identification of bad data in the power system. Aiming at the shortcomings of this method, a new data preprocessing method and clustering method are proposed. The improved GSA method is a technique to determine the most appropriate number of clusters. The original measurement data is tested by the improved RBF neural network, and the obtained results are clustered by the improved clustering method, and then the optimal number of clusters is judged by using the elbow criterion. Compared to the original algorithm, it can determine the number of clusters faster, greatly reduce the running time, and successfully identify the bad data in the power system.