Unlock instant, AI-driven research and patent intelligence for your innovation.

A Prediction Method of Oil Properties Based on Spectral Linear Representation

A technology of linear representation and prediction method, applied in the direction of measuring devices, material analysis through optical means, instruments, etc., can solve problems such as difficult to deal with prediction problems, insufficient use of input information, lack of input data, etc., to achieve oil property prediction value exact effect

Active Publication Date: 2021-04-27
SOUTHEAST UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these methods do not make full use of the input information, the data processing is too simple, and the lack of careful consideration of the input data makes it difficult to deal with more accurate prediction problems.
The near-infrared spectroscopy modeling problem has a strong nonlinearity, and its input data is near-infrared absorbance data in the wave number range of oil products, which contains a lot of information, which poses challenges to the traditional parametric and non-parametric models

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Prediction Method of Oil Properties Based on Spectral Linear Representation
  • A Prediction Method of Oil Properties Based on Spectral Linear Representation
  • A Prediction Method of Oil Properties Based on Spectral Linear Representation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0036] (1) Determine the basic parameters n and λ of the model

[0037] The range of the number of oil samples is set to n=10-100, and the traversal search is performed with a step size of 2. In order to speed up the search, and considering the fact that the Euclidean distance of the spectrum of oil samples is small, the regularization parameter λ takes logarithmic equal intervals, lg(λ)=-12~4, with a step size of 0.02. Next follow image 3 The procedure shown determines the basic parameters n and λ of the model.

[0038] Now follow figure 2 The flow shown, for the parameter combination n=10, λ=10 -12.00 Evaluate model performance:

[0039] Take the first sample in the oil sample library as the test sample S test = S 1 =(X 1 ,Y 1 ):

[0040] Absorbance data of the first sample in table 1

[0041] spectral point 1 2 3 … 208 Absorbance -0.0120 -0.0099 -0.0066 … -0.0796

[0042] Its property value Y 1 = 92.4. Other samples except the fir...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present invention specifically relates to a method for predicting properties of oil products based on spectral linear representation; performing principal component analysis on the near-infrared spectral data of the calibration set and test samples, and extracting the first k principal components in the score matrix obtained by principal component analysis to establish principal components space, and based on the Euclidean distance in the principal component space, find the oil samples in the n calibration sets closest to the test sample, which are called adjacent samples; calculate the near-infrared spectrum weight w of the adjacent sample; use the near-infrared spectrum weight w By weighting the property values ​​of adjacent samples, the property value prediction of the test sample is obtained. The invention predicts the test samples through the linear combination with specific weights, and combines the advantages of the parameter model and the non-parameter model.

Description

technical field [0001] The invention belongs to the field of oil product property detection in petrochemical industry, and in particular relates to a method for predicting oil product properties based on spectral linear expression. Background technique [0002] The traditional oil product evaluation method can provide detailed crude oil property data, but its operation is complicated and takes a long time, and it is difficult to meet the real-time requirements of oil product property analysis in the process of oil product processing. At present, the modeling technology based on near-infrared spectroscopy is becoming mature. These methods include multiple linear regression, local weighted regression, partial least squares, etc., and are widely used in the prediction of oil properties. Although these methods have begun to consider using the idea of ​​local modeling to deal with the nonlinearity existing in practical problems, the nature of the linear parameter model still limi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G01N21/359G01N21/3577
CPCG01N21/3577G01N21/359G06F18/24133G06F18/241
Inventor 焦一平费树岷陈夕松
Owner SOUTHEAST UNIV