The Application of Chemometric Methods in Near Infrared Spectroscopy Analysis

The near infrared spectroscopy analysis method is applied to measure the octane number of gasohol, which is used widely by automobile in China. Based on principal components regression (PCR), the prediction model is created. In order to save time in creating model and avoid unuseful data being used in model, the correlation coefficient between principal components and samples property is used. In order to improve the predicting precision, the software use one global model and some local models during predicting new sample. The samples in local model are dynamic and selected from the global model according to the set principle of dynamic local model creating method.