Strong reflection recognition and separation technology based on seismic multiwavelet decomposition has been widely used in seismic interpretive processing. This technology can remove the influence of strong shielding and improve the identification accuracy of concealed reservoir. Aiming at the issues of decomposition efficiency and accuracy of multi-wavelet decomposition technology, the instantaneous characteristics of seismic signal were introduced as the initial value, and local scanning parameters were reduced according to the relationship between wavelet parameters. A three-parameter fast wavelet decomposition technology was proposed to improve the accuracy and efficiency of wavelet decomposition. The relationship between formation thickness and strong shielding energy was analyzed through model forward modeling, and the relationship between formation thickness and separation coefficient was established. A new application idea of strong reflection separation with variable coefficient was explored. The new technology was applied to the actual seismic data, and the separation results weakened the influence of strong shielding, highlighted the effective signal of concealed reservoir, and improved the accuracy of reservoir prediction. The strong reflection separation process fully considered the relationship between coal seams with different thickness and separation coefficient, and the separation result was more consistent with the actual geological conditions than conventional methods.