NEURAL NETWORK MODELS OF KILN AND GRINDING MILL FOR ON-LINE PROCESS CONTROL OF CEMENT PLANT

摘要

Cement kiln and clinker grinding mills are two of the critical operations in cement manufacture. The quality parameters of the cement depend on the type and extend of mineral transformations occurring in the kiln as well as the particle size distribution achieved in the closed circuit grinding. Control of cement kilns has inherent difficulties because the quality parameters are not amenable for on line measurement. Though on line analyzers are available for the cement particle size analysis , the instruments are expensive and prone to operational difficulties. An attractive solution to this problem is the use of soft sensors bases on neural networks. In the soft sensors the quality variables are related to process measurements by a NN and estimated on line.