In order to solve the problem of low detection accuracy in the previous detection methods of compound attack activities in power monitoring network, a detection method of compound attack activities in power monitoring network based on improved D-S is proposed. The original data is normalized and denoised, and the improved D-S theory is used to extract more accurate and important features from the original features of the data and classify them, and then the network attack data detection model is constructed. By calculating the current state of the input data, the power monitoring network compound attack activity detection method is designed. The experimental results show that compared with the previous detection methods, the average bit error rate of the proposed detection method is only 3.2% in different network attack data sets, and the detection accuracy is higher.