基于深度学习的海上风电场网络异常检测

Studies of Anomaly Detection of Offshore Wind Farm Network based on Deep Learning

海上风电电站传输距离远,中途极易出现复杂环境干扰。文章提出采用深度学习算法进行海上风电电站的网络异常检测,对数据流量进行二维图像切分,采用inception-v3网络模型进行特征提取,利用最后三层的特征进行特征融合,最终采用softmax进行检测识别。测试结果表明,模型对海上风电网络识别精确度和检测率较高,且保持了较低的误警率。

In view of the long distance propagation of offshore wind power stations,complex environmental interference is likely to occur in the process.A deep learning algorithm was proposed for network anomaly detection of offshore wind power stations.Firstly,2 d image segmentation was conducted for data flow.Features were extracted by the inception-v3 network model.Then,features of the last three layers were used for fusion,and softmax were finally adopted for detection and recognition.The test results show that the model has high recognition accuracy and detection rate for offshore wind power network,and maintains a low false alarm rate.