Calibration analysis on boiling point of fatty alcohols with artificial neural network

Taking the boiling point of fatty alcohols for research object, the paper analyzes the relationship between the characteristic parameters of fatty alcohols and its boilling point with ANN and stepwise regression analysis. A ANN model for calibration analysis on point of fatty alcohols was created by using Wiener index, atomic position index, and molecular connectivity index as input variables and boiling point of fatty alcohols as output variable. Through training and optimum seeking, the idea network model was created, in which the model structure is 3:3:1, the fitting accuracy is 87.17% to 99.94% and the mean value is 97.58%. As a comparison, the fitting accuracy of regression model is 83.70% to 99.99% and the mean value is 96.82%. The results show that ANN model is better than the multivariate linear regression model. In multivariate calibration of chemical analysis, the combining use of ANN and stepwise regression analysis can improve the stability and practical applicability of the model. In performance analysis of the ANN model, the combining of visualization of qualitative analysis and the calculation precision of quantitative analysis is an effective method to overcome transition fitting data.