Study The Application Of Artificial Neural Network In Experiments Optimal Design Of Response Surface Methodology

This paper applies artificial neural network to optima design of theprocess conditions of Manchurian Dutchmanspipe Stem processed withalkali in response surface methodology. The optimal design of experimentis the optimization under the guidance of the thought of the optimaldesign of experiment. It’s scientific and rational to achieve the best resultand improve the quality of the scientific research achievements. Theoptimal design experiment can reasonable arrangements for variousexperimental factors and strictly control experimental error even theneffectively analyze experimental data. It obtains reliable data to the hiltusing less manpower and time. So it has been widely used.Response surface methodology has become an important means foroptimization design of experiments. The response surface methodologyincludes several aspects like experimental design, model fitting andprocess optimization. Its aim is to optimize the response. Artificial neuralnetwork is the mathematical model or computational model of imitatingthe structure and function of biological neural networks. Now artificialneural network has become an essential means of chemometrics to solvechemical problems. Study on the application of artificial neural networkoptimizes of the process conditions of Manchurian Dutchmanspipe Stemprocessed with alkali in response surface methodology. According to theexperimental data of the process conditions of ManchurianDutchmanspipe Stem processed with alkali determines the factors of theManchurian Dutchmanspipe Stem processed with alkali. They are theconcentration of NaHCO3, processing time,processing times and theirslevels. The evaluation index is the removal rate of total aristolochic acid.Then based on the experimental influence factors, levels and the responsevalue to establish response surface methodology. Then according to theorthogonal experimental data of the process conditions of Manchurian Dutchmanspipe Stem processed with alkali sets up BP and RBF artificialneural network models to predict the removal rate of total aristolochicacid. Last by response surface methodology design and the forecast valueof the removal rate of total aristolochic acid optimize the processconditions of Manchurian Dutchmanspipe Stem processed with alkali. Wecompare the RSM-BP and RSM-RBF models. Result shows that theoptimum process conditions were as follows:soaked3time with0.05mol/L NaHCO3solution,soaked for24hours. Under these conditionsthe removal rate of total aristolochic acid increased from72.51%to81.23%. RSM-RBF model makes the original design has a deepoptimization.Study the application artificial neural network of response surfacemethod process optimal design of experiments. Artificial neural networkcan quickly and accurately build model and response surface methodologycan system effectively analysis the optimal production conditions. Wewill use artificial neural network and response surfaces methodology inthe on experiment design for optimization. Using the artificial neuralnetworks in response surface methodology process optimize theexperimental design. Experimentuse can be optimization by theory. It canget high quality design to explore new technology in actual production. Itcan be used in many areas such as pharmaceutical chemistry, biologicalsciences, food science. It has some theoretical significance and practicalvalue for experimental design as well as provide new research ideas andmethods for the optimal design of experiment.