Method for predicting physical property data of oil product by using near infrared spectrum

Abstract

A method for predicting physical property data of an oil product by using a near infrared spectrum. The method comprises steps of: classifying oil products by properties; collecting a group of sample from each variety; determining a near infrared spectrum of each sample; determining physical property data of each sample by a routine method according to oil product varieties; combining determined sample near infrared spectrums side by side and establishing a database according to determined characteristic data; obtaining fitting spectrum by fitting database spectrums; calculating a fitting degree according to the fitting spectrum and a spectrum to be measured; comparing with a threshold; for a sample to be measured with complete spectrum fitting, database spectrum characteristics participated in fitting predicting characteristic data of the sample.

Description

technical field

[0001] The invention is a method for predicting oil physical property data by using near-infrared spectrum, specifically, it is a method for predicting oil physical property data from database sample data by establishing a database of known samples. Background technique

[0002] Due to the characteristics of modern molecular spectroscopy (near-infrared, infrared and Raman, etc.), which are fast, can measure multiple properties simultaneously, and are suitable for online non-destructive analysis, they have been increasingly used in petrochemical, agricultural and pharmaceutical fields, especially It plays an active and irreplaceable role in process analysis. However, these techniques need to use chemometric methods to establish a calibration model, and the calibration model has a certain scope of application, and it is not once and for all. If there is a large difference between the composition of the sample to be tested and the samples in the calibration set...

Images