This study aimed to model the minimum spouting velocity (U_ms) of vegetable biomasses in conical spouted beds including five biomasses. A statistical analysis of the literature correlations corroborated the lack of accurate models, since the average absolute relative errors (AAREs) exceeding 26%. Therefore, a new simple correlation was developed based on reliable data and the least square fitting method (LSFM) that improves the predictions for U_ms. Furthermore, three predictive methods were also designed based on intelligent approaches. The model developed based on the Gaussian Process Regression (GPR) provides the most accurate results with excellent AARE and R^2 values of 5.42% and 97.02%, respectively, in the testing step. The novel models were found to be applicable for vegetable biomasses of different shapes, as they excellently described the trends of U_ms under different conditions. Finally, the parameters of greatest influences on the performance of the models were discovered through a sensitivity analysis.
