Study on soft-sensing of mill material level based on data fusion in neural network

Aiming at the problem that the detection of the mill material level is not accurate by using conventional methods, this paper samples the parameters of the mill, include grinding sound signal, the pressure difference between import and export, and the temperature difference between import and export, combines the BP neural network, inosculates the sampling data through the multi-source data fusion method, achieves the soft-sensing of the mill material level. The actual measured data in the field shows this method has good metrical performance, in support of the enough training data, the fusion result is very closed to the set-value, so this method laid the foundation for optimal control of mill.