The early warning of electricity recovery risk needs to rely on a large number of data, including electricity consumption data, user information, historical arrears records, etc. However, the quality of these data may fluctuate and be unstable, which affects the accuracy and accuracy of the early warning model. Therefore, based on frequent item sets and big data, an active early-warning model of electricity bill recovery risk is constructed. Based on the power big data, the power rate recovery risk warning level constraints were determined, and the power rate recovery risk warning index weight was optimized. Considering frequent item set, the unified data model of electricity charge is constructed by distributed memory data grid technology. Through the fuzzy comprehensive evaluation, the electric charge recovery safety risk warning index is normalized to realize the active early warning of electricity charge recovery risk. The experimental results show that the model constructed in this study has a small deviation in active early warning of the risk of electricity bill recovery, and is practical.