Power grid DDoS detection method fusing multi-scale local and global information

The invention discloses a power grid DDoS detection method fusing multi-scale local and global information, and the method comprises the following steps: obtaining the characteristics of input flow, and converting the characteristics into a flow input sequence through an embedded layer; the traffic input sequence and the Embedding vector are input into a DDoS-MSCT model, and a CLS token is obtained; and obtaining a classification result of the input traffic according to the CLS token. According to the method, the local feature extraction module and the global feature extraction module are combined, so that the feature mode and rule of the DDoS attack are comprehensively captured. The local feature extraction module is composed of a convolution layer composed of a plurality of parallel convolution kernels of different sizes, convolution and pooling operation is carried out on input data of different scales, the multi-scale representation capability enables multi-scale convolution to better capture complex feat