Soft Sensor Modeling of Process Industrial Process with Uncertain Information

A soft-sensing model for key parameters of process industry driven by both qualitative and quantitative information is proposed, which aims at the characteristics that key parameters in process industry can not be measured online, multi-sub-processes, qualitative information and quantitative information coexist. For the sub-processes with quantitative information as the main part, a small amount of qualitative information in the process is quantified, and the sub-process model is established by data-based modeling method; for the sub-processes with qualitative information as the main part, a small amount of quantitative information in the process is quantified, and the sub-process model is established by knowledge-based modeling method. The soft sensing results of the final key parameters are obtained by using the model synthesis. Based on the hydrometallurgical process, a hybrid soft sensor model based on case-based reasoning (CBR) and RBF network in series is proposed. The simulation results show that the model is effective.