Abstract: In order to obtain the ideal results of the automatic identification of the illegal invasion of the Internet of things, an automatic identification method of the illegal invasion of the Internet of things based on the deep learning network is proposed. Firstly, we use the signal of Internet of things illegal invasion, and extract the time-domain and frequency-domain characteristics of abnormal invasion signal, then take the characteristics as the input of deep learning network, and the type of Internet of things illegal invasion is as the output. Through the training of deep learning network, we establish the Internet of things illegal invasion recognition classifier. Finally, we carry out the simulation experiment of Internet of things illegal invasion recognition with other methods The results show that the deep learning network can obtain high-precision pattern recognition results of Internet of things intrusion behavior, which can effectively ensure the security of Internet of things, and has a certain practical value.