Optimization Control Of Pulp Washing Process Based On Neural Network

Black liquor in the pulping process is one of the major pollution. Theeffective way to approach it is the high cost of alkali recover technology. Theway to reduce the cost is increase the concentration of black liquor. So, in thepulp washing process, it not only keeps the high quality of pulp washing, butalso requires obtain thick black liquor. Recently, the mechanism of pulp washinghas been extensively researched, but due to the unpredictable and multi-variabledisturbance, the research on equipment and process improvement is not enough.Automatic control and integrated optimization of pulp washing process shouldbe fully studied.This paper was supported by natural science foundation of china. In order tokeep pulp washing process work smoothly, this paper build the mathematicalmodel of the pulp washing process and design the optimization control system.The results of this work can be described as follows:(1) The soft sensor model of pulp washing process has been proposed.Based on the field data collected by DCS and simulation data obtained bymechanism, the soft sensor model of the residual soda and the baume degreewere obtained by two-step neural network and least square method identification.In addition, the neural network model was revised online. It shows that themodel has a high prediction.(2) Single objective and multi-objective optimization model have beenstudied. Based on material balance of washing and evaporation section, a singleobjective related to dilution factor and the cost of washing water has been build,in order to improve the actual production efficiency, multi-objective whichrespect to high quality, high yield and low consumption have been studied.(3) The predictive inference control of dilution factor has been proposed inthe article. As the series characteristic of dilution factor, plasma layer thicknessand washing water flow, establish the soft sensor model of plasma layerthickness. Based on this model and prediction inference control algorithm, dilution factor can be estimated. It shows that the algorithm improved theresponse speed of unpredictable variable, it also can be a monitored value forproduction process.(4) Optimization control system has been designed. Based on the hardwareof PLC of Siemens S7-400and software of WinCC6.0&Step7, a DCS withthree lays for pulp washing process was established. The basic level setsub-control systems of concentration, level, flow and temperature, which keepthe pulp washing process work smoothly. In the linkage level, small units linkedto adapt the changes of production. In the optimization level, mathematicalmodel of washing process can be identified and the indicators can meet theoptimization requirements as much as possible.