The shale gas exploration in Zhongmu area has a good prospect.However, the shale reservoirs in this area have the characteristics of strong heterogeneity, complex spatial distribution and vertical superposition.Due to the low accuracy of common inversion methods for thin reservoir lithology identification, the lithology identification problem in this area cannot be solved.Therefore, under the framework of frequency division inversion theory, this paper adds full-band information and explores a multi-wavelet frequency division inversion technique based on thin reservoir identification.Firstly, based on the theory of harmonic frequency extraction, the harmonic frequency extraction data is obtained, and the effect before and after denoising is compared. Secondly, the matching pursuit method is used to calculate the frequency division, and the high-precision frequency division data volume is obtained, which provides an important data basis for the subsequent heterogeneous characterization of thin reservoirs. The P-wave impedance-uranium-free gamma curve is reconstructed by rock physics relationship analysis to better distinguish shale. The low-frequency model obtained by Kriging interpolation algorithm, the medium-frequency model obtained by spectral inversion algorithm and the random high-frequency model based on wells are processed by frequency-division fusion, and a multi-scale fusion model suitable for thin reservoirs is established. Finally, the multi-wavelet frequency division inversion based on the constraint of reconstructed logging curve is carried out, and the effective identification of thin reservoirs is realized.The advantage of this method is that it can obtain a more accurate initial fusion model, highlighting the multi-wavelet and full-band information.