Abstract: In order to support the sharing of power IoT data and give full play to the value of power grid data, aiming at the problems of diverse technical components, difficulty in application, difficulty in finding data, high data application threshold, imperfect data model management and control mechanisms in the current power IoT data platform, this paper optimizes the overall framework of the power IoT data platform and proposes an isolated forest-based measurement real-time data quality anomaly detection and improved algorithm. The feasibility of the data quality detection algorithm is verified through two scenarios of experiment and emergency repair failure judgment and power failure judgement. It can provide a certain technical support for the construction of the power IoT data platform.