Abstract: The complex imperial sintering process possesses strong nonlinear, strong coupling, uncertain and multi-constrained features. In the paper, the neural network models are used to predict the permeability and burn through point(BTP), and the penalty function, parallel uneven grid method based on clustering and optimum maintaining simple genetic algorithm(OMSGA) are used to complete optimization of the whole status. The algorithm is globally convergent and industrially effective, resolves the real-time intelligent optimal control problem in complex imperial sintering process, and can be applied to practical industrial process with better results.