Prestack seismic inversion using prestack seismic gather data has become a commonly used technical method for reservoir prediction and fluid detection. In particular, for the exploration and development of unconventional oil and gas reservoirs such as tight reservoirs, shale reservoirs, the prediction and description of geophysical "sweet spot" reservoirs by using the prestack seismic inversion is an important topic and a difficult problem in research. The commonly used prestack seismic inversion method is a linearized approximation based on the exact Zoeppritz equations, and mainly uses the reflected compressional wave data. Considering that the approximate equation has the condition limitation of small incident angle in the prestack seismic inversion, it is not applicable to the far offset data, while the far offset data is conducive to improving the inversion accuracy of the elastic parameters. There is an error between the approximate equations and the exact Zoeppritz equations during forward modelling, and the accumulated error during the inversion process will reduce the inversion accuracy, so the use of the exact Zoeppritz equations can effectively improve the inversion accuracy. In order to improve the suitability of prestack seismic inversion, an inversion method using the exact Zoeppritz analytical expression is proposed, avoiding the introduction of condition constraints which are approximated by the exact Zoeppritz equations and greatly improving the accuracy of prestack seismic inversion. Moreover, the reflected compressional wave data and the converted wave data are used in the prestack seismic inversion. The inversion of reflected compressional wave data and converted wave data is combined to improve the accuracy of the inversion result by increasing the input information in the prestack seismic inversion. Besides, the joint inversion of the reflected compressional wave data and the converted wave data based on the exact Zoeppritz equations can obtain more accurate parameters of compressional wave velocity, shear wave velocity and density, which can better describe the reservoir, detect the fluid, and determine the favorable "sweet spot" area of the oil and gas reservoirs . Furthermore, the Iteratively Regularized Levenberg Marquardt method (IRLM) is used in the inversion calculation process to solve the strongly nonlinear and ill-posed problems of multi-parameter inversion before stacking, and improve the robustness of the IRLM inversion method. Finally, through the application of synthetic data and actual data, the inversion results verify the applicability of the IRLM inversion method, and the method has strong robustness and can be applied to the actual data.