Abstract In view of the security threats posed inside and outside the network due to the weaknesses of network protocols and systems, power line and wireless converged communications have security problems in communication connections that appear in environments where heterogeneous networks overlap. In order to cope with the problems of network delay and data transmission security in smart grid, this paper proposes an online monitoring security protection method based on edge nodes, which manages user trust under the premise of not leaking grid user data, and establishes a personalized machine learning model. This method not only guarantees users a secure communication experience, but also ensures users’ real-time network connection and content acquisition, and can obtain a fast and personalized user experience.