Abstract: With the development of new power distribution networks, energy storage devices have been deployed in low-voltage distribution networks. The energy storage converter is the interface of energy storage and power grid whose switch is a fault-prone device, and the open circuit fault is not easy to find. Aiming at the complex power quality environment of distribution grid on the user side, based on the mechanism of fault analysis, this paper proposes an open circuit fault diagnosis method of switches based on extended Kalman filter (EKF) and broad learning system (BLS). The method uses EKF to extract the feature of fault harmonics and BLS to identify features and diagnose faults without manually setting the threshold. Through simulations and experiments, the fault switches can be accurately located under the common power quality disturbances in distribution grid such as load fluctuation, three-phase imbalance, harmonics and frequency drift, which effectively provides the supports for monitoring the energy storage system on the user side.