| Simulated moving bed (SMB) chromatographic separation technology is widelydeveloped in the 60's last century. It possesses all the good aspects such as greater separationcapacity, smaller bulk, lower costs, and is easier to control and apply in large scale. All theseadvantages make it the most important position in chromatographic technology. With thedevelopment of industrial automation, people pay more and more attentions on higherproductivity, higher product quality and lower cost than just make devices stable moving.Measurement of component purities by soft sensor technology and SMB separationprocess stable optimization designs and optimal algorithms are intensively studied here. Allstudies are for on-line optimization and control in industrial field and exactly supply valuablereference.There are two methods for component purities measurement recently, the first one isdone by manual analyze with long time lag which cannot warrant smooth optimization andcontrol, the other one is using on-line measuring instrument with high cost. So instrumentwith low cost and good use is needed urgently. A soft sensor of purities measurement isrealized by BP neural network. Two kinds of modeling are used, stationary-state modelingand dynamic-state modeling. To compare with dynamic-state modeling, Method ofstationary-state modeling is simpler and faster, but need more accumulated data.Optimization methods under stationary-state separation process are also mainly studied.Two optimization designs are introduced firstly under both traditional graphic methodtriangle theory and recent studies, one is based on variable m, the other β. GA and SQP areapplied respectively to these two designs and much simulation is done to validate them. In theapplication of GA, a method is given to make GA easier when GA with nonlinear constraintsturns into GA without constraints. To compare with GA, SQP is more convenient and fastwhen they are used in SMB stationary optimization. |
PDF Full Text Request