On the basis of summarizing the advantages and disadvantages of traditional road noise measurement methods, this paper draws lessons from the measurement method of tyre-pavement coupling noise of voters, an embedded traffic roaming patrol sensing system developed by Professor Ming L. Wang of Northeastern University, and optimizes the position parameters of the microphone. Through the fitting research of driving speed and equivalent continuous A-weighted sound pressure level, the equivalent continuous A-weighted sound pressure level at different speeds can be converted into the equivalent continuous A-weighted sound pressure level at standard speeds, and the speed correction of the equivalent continuous A-weighted sound pressure level can be realized. The spectral subtraction weight algorithm is used to suppress noise interference, the denoised signal is windowed and reconstructed, the tire road coupling noise characteristic signal is extracted by the principal component analysis method, and the envelope noise energy (ENE) is obtained by integrating the extracted first principal component (F1) in the frequency band of 100Hz ~ 1kHz. Data fitting of ENE and SFC under different test sections is carried out by SPSS mathematical statistics software. The goodness of fit of ENE and SFC is the highest under the power function model, and the corresponding goodness of fit R2 is as high as 0.98, which can provide a new direction for the detection technology of pavement skid resistance performance.