POWER IOT INTRUSION DETECTION METHOD BASED ON SSAE AND IMPROVED INDRNN

With the continuous integration of internet of things (IoT) technology and power system, there are endless intrusions to power system launched by IoT terminal devices. In order to improve the protection ability, this paper proposes a hybrid intrusion detection model based on stacked sparse auto-encoder (SSAE) and independently recurrent neural network (IndRNN). SSAE was used to solve the problem of large number of redundant features in high-dimensional data of the power IoT, and the improved IndRNN was used to capture timing information and introduce hierarchical attention mechanism to enhance key features. Experimental results show that the accuracy rate and false positive rate reach 99.36% and 0.67%, and it greatly shortens the detection time, which is an effective intrusion detection model of power IoT.