Hybrid Modeling of Catalytic Cracking Processes Based on the 17-Lump Kinetic Model and Light Gradient Boosting Machine

Fluid catalytic cracking is a crucial procedure in the petrochemical industry, but understanding and predicting its process remains challenging due to its high degree of complexity and nonlinearities. This paper proposes a hybrid model incorporating a 17-lump kinetic model (SLKM) and a light gradient boosting machine (LightGBM). First, as the difficulty of parameter determination of SLKM, this paper improves the dung beetle optimization algorithm for solving the parameters in SLKM. The algorithm simulates the search behavior of insects and uses a fitness function to optimize the parameters, thereby addressing the inaccuracy and time-consuming nature of parameter solving in the mechanistic model. Second, the challenge of hyperparameter optimization in LightGBM, which greatly influences the precision of model predictions, is addressed in this paper by applying Bayesian optimization with Hyperband. The prediction performance of the LightGBM model improves significantly following this hyperparameter optimization. SLKM, backed by existing theoretical knowledge, produces outputs consistent with physicochemical laws and requires less data, though its prediction performance is limited. To enhance performance, we construct three hybrid models by combining SLKM and LightGBM: the series model, the parallel model, and the hybrid model with adaptive weights, realizing the complementarity of both models. The industrial data validation results demonstrate that the adaptive weighted hybrid model achieves the best prediction performance, with the mean relative error, mean squared error, and mean absolute error reaching 1.36%, 0.0169, and 0.077 on the industrial plant data set, respectively.