Intelligent Recognition and Classification of IoT Devices via Information Physics-Based Multi-Source Data Association

With the development of information and intelligence in the power system, the massive integration of heterogeneous IoT devices poses new challenges in system administration and security. To address the classification and recognition challenges of IoT devices in power system, this paper proposes an intelligent approach that integrates multiple data sources with logical modeling. This method models the communication data of IoT devices as state transition graphs, augments them with statistical information to obtain attributed graphs, and utilizes graph kernel methods to measure the similarity between graphs, thereby achieving the classification and identification of IoT terminal devices. To reduce computational costs, local sensitive hashing (LSH) technology is introduced to maintain the attribute information of the original data. The proposed method aims to enhance the accuracy of IoT device identification and the efficiency of classification in the power system, providing robust support for system management and security maintenance.