To improve the prediction accuracy of the carbon content, a predictive model based on the improved BP neural network(BPNN) is presented by using the modified teaching-learning-based optimization(MTLBO) algorithm. In order to improve the global optimization ability of the basis teaching-learning-based optimization(TLBO) algorithm, a preview phase and a review phase are added in the MTLBO algorithm. The quantum-behaved learning strategy is adopted in the learning phase. Test results show that the MTLBO algorithm is valid to solve the global and local optimization problems. The MTLBO algorithm is used to optimize the weights and thresholds of the BPNN for the prediction of catalyst carbon content of the catalystic reforming unit. The simulation results show that the predictive precision and generalization ability of the proposed method have a certain degree of improvement.