Smart grid DoS attack detection method based on DQN algorithm

The invention discloses an intelligent power grid DoS attack detection method based on a DQN algorithm. The traditional bad data detection method is very dependent on threshold setting, and the degree of dependence on the threshold setting is reduced by using the DQN algorithm. In the aspect of observation value construction, the observation value is constructed in a continuous product mode, and compared with a traditional sliding window method, the dimension of the observation value can be greatly reduced. Meanwhile, after the attack is detected, unreceived signals are rejected to obtain a new state estimation, and the estimation can be used as a reference for subsequent calculation. And finally training DoS attack detection by using a DQN algorithm. According to the method, the DoS attack is rapidly detected, the detection speed and the success rate are increased, the implementation method is simple, and the detection efficiency can be remarkably improved. 本发明公开了一种基于DQN算法的智能电网DoS攻击检测方法。传统的坏数据检测法十分依赖阈值设置，本发明