BPNN and RBFNN based modeling analysis and comparison for cement calcination process

Abstract

In order to improve the production stability of cement Precalciner Kiln calcination process, it is necessary to conduct in-depth analysis of the calcination process, knowledge of the process in running state and laws. To save energy and achieve stable production, we establish the simulation model of the calcination process used to find effective control methods. In view of the calcination process parameters of complex mathematical model is difficult, so we expressed directly using neural network method to establish the simulation model of the calcination process. Choosing reasonable state and control variables and collecting actual operation data to train neural network weights. Constructed two types of neural network BPNN and RBFNN based models, both achieved good fitting results. RBFNN based model can reach very high fitting results, but the BPNN based model has good generalization ability. So the BPNN based model can be used as simulation model of the calcination process for exploring new control algorithms.