Aiming at the problem that the current electric pump well production operation alarm system cannot solve the problem of oil well danger degree ranking, the fuzzy comprehensive evaluation method is used to carry out the online pre-alarm model and application research of electric pump well based on real-time monitoring data. Based on the monitoring data of the electric pump downhole and wellhead, the collected 11 working condition parameters are divided into four categories according to their types: electric parameters, pressure, temperature, and production. A second-level pre-alarm index system for electric pump wells is established, and the weight vector of each index is calculated by using the analytic hierarchy process(AHP). The membership function of each secondary index is established based on expert experience and actual working conditions, and the membership matrix of each index is obtained. The calculation model of the electric pump well pre-alarm index is obtained by multiplying the weight vector and the membership matrix, and the on-line pre-alarm software of the electric pump well production operation is compiled. Field application shows that the pre-alarm index model and software can sort oil well danger levels in real time, accurately and quantitatively, solve the problem of timely and accurate alarms for safe operation of electric pump wells, and improve the intelligent management level of electric pump wells.