Abstract: To meet the requirements of the ultra-high speed protection on UHV transmission lines, this paper proposes a new starting algorithm based on wavelet transform and neural network. This algorithm uses the principle that the wavelet transform can detect mutations in the signal and extracts the features of fault signals as the inputs and designs a starting neural network, which can judge the line running state from the output to determine whether start the fault phase selector. By simulating UHV transmission line system in PSCAD, it is proved that the proposed algorithm can detect the fault accurately and output the reliable action quickly. It's also not influenced by fault resistance, fault location, fault inception angle and noise interference. It has a practical value for ultra-high speed protection of UHV transmission lines.