Automatic feature extraction model of power iot terminal

The feature extraction technology of Internet of Things terminal is the focus of current research. In this paper, an automatic feature extraction method of power Internet of Things terminal is proposed, which includes: identifying the flow information of power Internet of Things terminal; Set feature classification rules to classify and label device feature information into shallow feature and deep feature. The shallow feature of the device is extracted from the edge node, and the deep feature of the device is extracted from the cloud. Analyze the abnormal characteristics of each label, analyze the correlation degree of extracted features, and screen out the feature values with high correlation degree; Construct multi-mode feature extraction model; The feature extraction model of multiple patterns is constructed by combining workflow automation orchestration. Based on packet resolution, traffic uploaded by devices is detected and classified. Based on actual service conditions, different feature information is extracted on edge terminals and the cloud cloud as required, and the feature extraction process is automated to improve computing efficiency, reduce network delay, and reduce cloud computing and storage costs.