Based On The Matlab Rhamnoiipid Fermentation Parameters Optimization Design And Implementation

Rhamnolipid is the metabolism,and one of the biosurfactant that produced bypseudomonas or burke’s fungus,and also a kind of biosurfactant that was researched、applicated for the longest time.The daqing victex chemical Co., LTD. On a lot of the scientific research results of the[1,2,3,4,5]digestion and absorption in2006after the introduction of rhamnolipid fermentationpilot technology and start industrialized production of exploration, and in the following yearbuilt12000tons/year fermented liquid production line, the production rate of25to35g/Lrelatively stable level, individual cases also produced the40g/L products, and Americanindustrialization device rat lee glycolipids biological fermented liquid tank under the samelevel[6]. But also there are often yield level very low case, and with the weather environment,culture medium formula basic has nothing to do, posts and all operation parameters in processcontrol standard interval.The rhamnolipid fermentation production operation is fat by boiler, air compressor,ingredients for materials, fermentation tank temperature control, laboratory testing fiveposition to finish the operators, four team3, a team seven operator, each post process standardalso made very science, each post should be strictly according to post process operation shallcarry out the work of responsible. But the technical personnel of all operation parameters,production results are compiling, analysis, do so and there is no guarantee that yield on ahigher the stable value, think in many operating parameters process standard interval, theremust be a the most reasonable operation parameters, that is to say, this is a typicalmulti-objective optimization problem.Using artificial neural network to nonlinear multi-objective problems associatedoptimization algorithm is feasible scheme[7,8]This article through to the discussion of theartificial neural network, the forward of the feedback of artificial neural network (BP network)build a8-10-1structure of the rhamnolipid fermentation parameter prediction operation theyield of model, with random parallel methods select accounted for three-quarters of thestatistical data of operation parameters data model for training, at the same time use thegenetic algorithm to the BP network weights and threshold for a whole domain optimization, then use another one quarter of operating parameter model simulation, the results andstatistical operating results production rate were compared, the error of the reaching set value.In the operating parameters optimization and production rate of the biggest prediction results,in order to reduce the workload operation model, this paper has not adopted according to themethod of genetic algorithm to eight operation parameters for global optimization scheme, butthe scatterplot statistical methods, according to the operation parameters and the yield ofrelational sequence, the selection of the production rate and the correlation of the largest twoparameters are optimized. And also use scatterplot statistics method, the optimization of thefirst to target operation optimization of parameters interval out, then press production linegeneral operation method, for the selected two parameters optimization of the optimal schemeinterval, specific gradient setting. In order to avoid scatterplot statistical method scatterplotconcentration distribution area lay outside the corresponding points of the misjudgment, andin the scatterplot statistical figure, press production rate of the larger the principle ofconcentration distribution points, and the most high in foreign also choose scatterplot relativeconcentration of two to three points, the selected parameters optimization areas added. Otheroperating parameters according to the scatterplot statistics choose scatterplot of the mostconcentrated parameter values. Then orthogonal method selected operation parameterscombination, gets63group operating parameters, so that they will be in eight type parametersof tens of thousands of parameter combination in the problem of optimal, simplified to twoparameters in a limited area of the63group of parameters optimization problem.In using the above methods to get maximum production rate and the correspondingoptimization parameters combination, from fermentation theoretical Angle, to other operatingparameters change direction and the production rate of change of direction for the correlationanalysis, the choice in addition to the above two operating parameters outside, theory andproduction rate is relatively large connection of the three parameters, and combining with thedirection of the production rate of increase in theory with scatterplots statistical results, butagain not beyond the principle of process standard, to the three parameters is revised,respectively with the above two parameters with BP neural network model for solving themaximum production rate, from the operation result see, revised production rate than themaximum value solution before solving maximum value increased by10.84%.