Abstract: To address the low accuracy of traditional rolling force calculation methods, this paper proposes a PSO-LSSVM-AdaBoost model (Particle Swarm Optimization-Least Squares Support Vector Machine-Adaptive Boosting) based on machine learning to predict the rolling force of horizontal in hot rolling of H-beam. First, dimensional parameters of H-beams with various specifications, equipment parameters, and relevant rolling parameters were collected to create a multi-feature rolling force dataset. Outliers were removed using the Isolation Forest (IF) algorithm. Next, the Least Squares Support Vector Machine (LSSVM) parameters were optimized using the Particle Swarm Optimization (PSO) algorithm, while the weight assignment mechanism of the adaptive AdaBoost algorithm was employed to further enhance the model's predictive performance. Comparisons with traditional Support Vector Machine (SVM), Particle Swarm Optimization-Support Vector Machine(PSO-SVM), Least Squares Support Vector Machine(LSSVM), Particle Swarm Optimization-Least Squares Support Vector Machine(PSO-LSSVM), Least Squares Support Vector Machine-Adaptive Boosting(LSSVM-AdaBoost)models showed that the proposed model achieved greater accuracy and stability in predicting the horizontal rolling force of H-beam, can provide a reference for hot-rolled H-beam production.