Accurate and intelligent leakage localization of the urban directly buried heating pipe network is the key aspect for the urban heating infrastructure safety, energy-saving and carbon-reducing and intelligence operation. In order to improve the localization accuracy of the direct buried hot water heating pipe network leakage by the acoustic wave method, a DN300 diameter large scaled heating experiment system was established. The effects of acoustic wave propagation distances, temperature, pressure and flow rates on the acoustic wave signal characteristics with and without leakage were studied. The acoustic wave time-frequency characteristics from pipeline pump vibration were analyzed, and then the wavelet threshold method was used for signal noise reduction. It obtained the leakage characteristic frequency band range. The results show that temperature has significant effect on the acoustic frequency band of leakage, and higher temperature contributes to larger acoustic wave frequency band. When the temperature rises from 31 ℃ to 86 ℃, the characteristic frequency band range expand from 200 ~ 800 Hz to 50 ~ 1 500 Hz. The propagation distance, pressure and flow rate mainly affect the amplitude of leakage signal, but they have little effect on the band range. The acoustic energy of the pump is mainly within 3 000 ~4 000 Hz, and it is beyond the range of the leakage characteristic frequency band range. The wavelet threshold method for the signal noise reduction. It could improve the localization accuracy up to 0.11%, along with a localization distance deviation within ± 1 m.