Data-driven integrated modeling and intelligent control methods of grinding process

Abstract

The grinding process is a typical complex nonlinear multivariable process with strongly coupling and large time delays. Based on the data-driven modeling theory, the integrated modeling and intelligent control method of grinding process is carried out in the paper, which includes the soft-sensor model of the key technology indicators (grinding granularity and mill discharge rate) based on wavelet neural network optimized by the improved shuffled frog leaping algorithm (ISFLA), the optimized set-point model utilizing case-based reasoning and the self-tuning PID decoupling controller. Simulation results and industrial application experiments clearly show the feasibility and effectiveness of control methods and satisfy the real-time control requirements of the grinding process. © 2012 Springer-Verlag.