A new fluid catalytic cracking (FCC) lumped kinetic modeling method was put forward to improve the prediction accuracy of the FCC kinetic model. This model divided the feedstock into three lumps and the products into six lumps. The parameters of the kinetic model were obtained by combining the fourth-order Runge-Kutta integral method and the two-swarm cooperative particle swarm optimization algorithm. Compared with dividing the feedstock according to the group composition and the distillation range, the average relative error between the experimental data and the calculated values proves the higher fitting accuracy of the new lumped method. The kinetic model established by the method exhibits better adaptability to different feedstock and fewer requirements for feedstock analysis data.