Abstract: Based on the tropical regional atmospheric model system of China Meteorological Administration (CMA-TRAMS), the forecast of tropical cyclone (TC) formation for the No. 3 Typhoon Chaba in 2022 are analyzed, as well as the comparison of prediction between CMA-TRAMS and European Centre for Medium-Range Weather Forecasts (ECMWF). The reasons of the successful genesis prediction of CMA-TRAMS for Typhoon Chaba are investigated from different aspects such as the development environment of tropical cyclone (TC) embryos, physical processes within TC embryos, embryo structure and the corresponding development. The results indicate that numerical models are required to have good descriptive ability in the above aspects to perform well in TC generation forecast. This study is beneficial for us to understand the main physical processes and factors closely related to TC generation prediction in numerical models, and it would provide clues for the follow-up model development and improvement. The results indicate that the significant difference in the cyclonic circulation pattern of the monsoon trough in the western Philippines between the 72-120 h forecasts is the direct cause of the difference in TC generation forecasts between the two models. CMA-TRAMS predicts the generation, development, and merging of multiple mesoscale convective systems (MCSs) or mesoscale convective vortices (MCVs) in a positive vorticity environment between 96-120 h, and the organization of circulation to form warm core structures, which is important in its successful prediction of the formation of "Chaba The accuracy of wind field forecasting in the western monsoon trough of the Philippines may have a significant impact on the prediction of TC formation in the South China Sea. The continuous convergence and merging of MCSs and MCVs, and the organization of cyclone circulation are important physical processes for TC formation. The results of this study have deepened our understanding of the main physical processes involved in TC formation, enhanced our knowledge of the influencing factors of numerical model prediction for TC formation, and provided clues for improving model forecasts.