It is required to conduct repeated multi-frequency test to determine the best detection frequency when taking traditional tone burst as the guided wave excitation of dirt pipe, which had problems of blindness, time and resources consuming. A conversion algorithm from broadband chirp to tone burst was put forward based on broadband chirp guided wave excitation of dirt pipe. All tone burst responses in the broadband of chirp signal is similarly extracted during post-processing. The best detection frequency was determined through the comparing energy distribution and modal purity of the different frequency guided wave responses that it was calculated from the chirp response, and combined with time-frequency analysis method to determine the appropriate waveguide frequency interval according to the distribution of the response waveform energy of the dirt pipe. It had shown from simulation analysis and experiment studies that: the center frequency of about 175 kHz was the most appropriate for guided wave detection of investigating dirt pipe. The method could determine the best guided wave detection frequency of dirt addresses both test at multi-frequency and minimize acquisition time, which would lay a foundation for achieving guided wave detection from the ‘point’ to ‘line’ of the pipe dirt.