If bad data exists in the leverage point of the power system, it will seriously affect the bad data identification and state estimation results. It is difficult for the existing robust estimation methods to effectively deal with the leverage measurement. The removal of the leverage point measurement may affect the observability of the system. This method can improve the bad data of the leverage point by transformation of the Jacobian matrix based on the bilinear WLAV state estimation. This method improves the algorithm itself. It does not need to eliminate the leverage measurements, which can not only improve the ability to resist the leverage point bad data, but also can not affect the observability of the system. It’s an ideal way to deal with bad data at leverage points. Simulation results based on the IEEE standard system and a real provincial network verify the effectiveness of the proposed method in improving the computation accuracy and computing efficiency. This work is supported by National Natural Science Foundation of China (No.51277052).