Abstract: The operation of the Xiaolangdi Reservoir and the improvement of river regulation works since 2000 have contributed to significant morphological changes in the medium-flow channel (main channel) of the Lower Yellow River. A characterization of the morphological adjustment of main channel cross-sections is essential for improving the understanding of regional riverbed evolution. This study analyzed the adjustment patterns of main channel cross-sections in the Baihe—Sunkou reach based on approximately 2 500 post-flood measured cross-sectional profiles from 2000 to 2021. A model for predicting the trends in main channel cross-section adjustments was developed by integrating Particle Swarm Optimization (PSO) with Support Vector Regression (SVR). The results indicate: ① At the cross-sectional scale, the main channel exhibited four types of adjustment patterns: widening with aggradation; widening with degradation, which was the dominant pattern; aggradation only; and degradation only. Temporally, the proportion of widening with degradation gradually decreased, while that of degradation only became more prevalent. Spatially, adjustment patterns were more complex in the Baihe—Huayuankou and Huayuankou—Jiahetan reaches. ② At the reach-scale, three adjustment patterns of main channel cross-sections were observed: widening with aggradation; narrowing with degradation; and widening with degradation, which was the dominant pattern. The contributions of lateral widening and vertical degradation to the expansion of bankfull area varied considerably across the four sub-reaches: 41%—59% in Baihe—Huayuankou reach, 37%—63% in Huayuankou—Jiahetan reach, 43%—57% in Jiahetan—Gaocun reach, and 22%—78% in Gaocun—Sunkou reach. ③ The error of the PSO-SVR machine learning model in predicting cumulative changes in bankfull width and riverbed elevation is < 9%, achieving an accuracy in predicting main channel cross-section adjustment trends exceeding 80%.