Abstract: Aiming at the problem of low detection accuracy for abnormal behavior of network states, we proposed an anomaly detection model  of network state based on data mining. Firstly, the network state signal was extracted, and the signal was pretreated by wavelet transform, and the features of the network
 anomaly detection were extracted. Secondly, the network state anomaly detection model was built by echo state network, and genetic algorithm was used to optimize the parameters of the echo state network. Finally, the network state anomaly data sets were used to test the effectiveness of the model. The test results show that data mining technology can accurately detect abnormal behavior of various network states.

Key words:  network anomaly, intrusion behavior, detection model, data mining