【摘要】 为保证船用通信网络的安全，设计基于深度数据挖掘的船用通信网络异常行为分类和识别方法。该方法数据处理模块采用全局信息数据融合策略，融合网络的原始采集数据，特征选择模块通过平均不纯度减少特征重要度计算方法，选择有效特征并计算该特征重要度后，形成特征集，将其输入分类识别模块的内外卷积网络深度学习网络模型中，通过模型的学习和训练，获取船用通信网络异常行为分类识别结果。测试结果显示：该方法可有效删除其中的无效特征，保留有效特征结果；可获取不同类别有效特征标签的重要度评分结果；分类识别的平均绝对误差均低于0.18，可完成不同流量变化下异常行为分类识别。更多还原

【Abstract】 In order to ensure the secure communication of marine communication networks and real-time safety of ships, a classification and recognition method for abnormal behavior in marine communication networks based on deep data mining is studied. The data processing module of this method adopts a global information data fusion strategy to fuse the original collected data of the network. The feature selection module reduces the importance of the feature reading calculation method by average impure. After selecting effective features and calculating their importance, a feature set is formed, which is input into the deep learning network model of the internal and external convolutional network of the classification and recognition module. Through model learning and training, Obtain the classification and recognition results of abnormal behavior in marine communication networks. The test results show that this method can effectively delete invalid features and retain valid feature results. The importance rating results of effective feature labels in different categories can be obtained. The average absolute error of classification recognition is less than 0.18. It can complete the classification and recognition of abnormal behavior under different traffic changes.更多还原