UHVDC transmission system has strong nonlinearity. The traveling wave ranging principle of UHVDC transmission line adopting wavelet transform method has adaptive problems in practical engineering. At present, traveling wave ranging technology has traveling wave arrival time and traveling wave velocity. It is difficult to organically unify. Aiming at this problem, a DC line fault location method considering the strong nonlinear system of UHVDC transmission and the variation of wave velocity is proposed. From the adaptability of the ranging method, an improved Hilbert-Huang algorithm which is very suitable for transient information processing is proposed. The algorithm can accurately calibrate the initial traveling wavehead of the fault. From the perspective of ranging accuracy, the wave velocity variation characteristics of the fault traveling wave are analyzed. It is found that the frequency variation characteristics of the line parameters and the attenuation of the traveling wavehead cause the traveling wave velocity to decrease with the increase of the fault distance. A neural network algorithm is proposed, which can achieve accurate fault location without calculating the traveling wave velocity. A large number of simulation results show that the ranging method has higher ranging accuracy and good robustness under different fault distances and different transition resistances. This work is supported by National Natural Science Foundation of China (No. 51767023) and Joint Natural Science Foundation of the Xinjiang Uygur Autonomous Region (No. 2017D01C028).