A technique based on Coiflets wavelet analysis for real-timely detecting and locating transient surges of network voltage is proposed. Such a wavelet method holds both the ability in the local time-frequency multi-resolution analysis and the fast algorithm for general wavelet method. Most interesting, the top-level scale coefficient is exactly the single-point sampling value of voltage signal, with no need for numerical integration required in general wavelet method. This operation can effectively reduce the calculation amount, save the consuming time, and improve real-time performance. By quantitatively studying four typical kinds of transient surges of network voltage, the results demonstrate that the proposed wavelet method is capable of accurately identifying various transient surges of network voltage, and the relative error of location can reach one thousand which is obviously better than that of existing methods. More important, this accuracy of location is insensitive to the type and amplitude of voltage surge, leading to an excellent reliability of analysis result in the practical application. This work is supported by National Natural Science Foundation of China (No. 11502103).