An accurate and reliable fault location method can ensure quick recovery of a fault line and improve the reliability of power supply in a hybrid three-terminal DC transmission system. In order to solve the problems of complex structure and difficult fault location of such an HVDC transmission system, a location method based on a wavelet packet energy spectrum combined with a BP neural network is proposed. The specific positioning strategies are as follows: first, the fault line is quickly selected when the fault occurs. Then through wavelet packet decomposition and reconstruction, the voltage component collected at the measurement point is converted into wavelet packet energy when the fault occurs. This is used as an input sample to train the nonlinear fitting ability of BP. Finally, the corresponding fault distance can be output by substituting the energy of the wavelet packet which reflects the fault position. From the simulation results, it can be seen that the proposed method is strong in anti-transition-resistance with a high accuracy in fault location. This work is supported by the National Natural Science Foundation of China (No. 51777166) and Key Support Projects of the National Natural Science Foundation of China (No. U1766209) and China Scholarship Council Fund.