Detection Method for Abnormal Flow Data in Power Business Based on Deep Learning

Abstract

As the scale and complexity of China's power information network continue to expand, various application services based on this network are becoming increasingly widespread, leading to a gradual increase in the volume of generated data. Consequently, the detection of anomalous traffic data is crucial for ensuring the stability and security of power operations. Traditional anomaly detection methods primarily rely on shallow machine learning algorithms, which have significant limitations in feature extraction and pattern recognition, particularly when dealing with high-dimensional, nonlinear, and complex temporal data. This paper introduces a deep learning approach for anomaly identification in traffic data, combining Bidirectional Long Short-Term Memory (BiLSTM) networks using the Whale Optimization Algorithm (WOA) to achieve superior detection precision and robustness. Experimental findings reveal that the proposed method markedly enhances both accuracy and efficiency in comparison to conventional techniques, offering a more effective solution for anomaly detection in power business operations.