Fault Classification Method of Information System Based on Cascaded Convolutional Neural Network Model

Artificial intelligence plays an important role in data anomaly detection, which can significantly improve the speed and accuracy of fault diagnosis and detection in power information system. With its powerful fitting ability, deep learning begins to emerge in various fields, especially in the field of da-ta anomaly detection, which can also use deep learning methods. Due to various types and similar features of power data sets, fault classification cannot be carried out accurately. At present, frequent detection of text information needs to be performed manually. To solve this problem, this paper proposes a fault classification method for information system based on cascaded convolutional neural network model. BERT deep learning network is used for model training, and a cascaded classification network is constructed. Multiple BERT models are called hierarchically to achieve the fault detection effect of hierarchical classification. The method presented in this paper improves the accuracy of fault prediction and classification, which also provides fast, accurate and reliable decision information basis for improving fault diagnosis.