Industrial control network abnormal flow detection method based on data enhancement and deep learning

The invention discloses an industrial control network abnormal flow detection method based on data enhancement and deep learning, and relates to an industrial control system safety detection method. According to the model, a hybrid sampling algorithm (ADASEN) is used for balancing data, the algorithm fuses an adaptive oversampling algorithm (ADASYN) and a repeated editing neighbor rule undersampling algorithm (RENN), density distribution information of samples is fully utilized, decision boundary samples are concerned, and the problem of data imbalance is effectively solved; the data feature importance is calculated by using a Gain method of an XGBoost algorithm, the Pearson correlation analysis is introduced to calculate the correlation coefficient between the features to optimize the data feature importance, a deep cross neural network model is constructed, the data features are fully extracted to improve the anomaly detection accuracy, and the anomaly traffic detection is completed. According to the method