Optimization Study of Setting Reference Values based on Combined Cement Grinding System

In order to mitigate excessive fluctuations in control indicators and ensure stable performance of cement products, this paper proposes a research method for optimizing setting reference values in a cement combined grinding system. Based on an analysis of the cement combined grinding process, historical production data is utilized to establish a predictive model for process indicators using particle swarm optimization with extreme learning machine. Subsequently, an objective function is established, and in combination with the predictive correlation model, the differential evolution particle swarm algorithm is used to offline compute the set reference values for the rotational speed of the circulating fan and the rotational speed of the powder separator. Simulation results validate the effectiveness of this method in establishing setpoint values for process indicators in cement combined grinding, laying a foundation for stable control in practical production processes.