Rapeseed quality detection method based on characteristic wavelength optimization and multi-model fusion

By using a method of characteristic wavelength optimization and multi-model fusion, the problems of high hardware cost, data redundancy and poor adaptability of multiple indicators in rapeseed quality testing are solved, realizing high-precision, rapid and non-destructive testing of multiple indicators of rapeseed, which is suitable for large-scale testing in the field.

CN122084567APending Publication Date: 2026-05-26HUAZHONG AGRI UNIV
0 Cites 0 Cited by

Patent Information

Application Number
CN202511899496.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-16
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing technologies for rapeseed quality testing suffer from problems such as high hardware costs, data redundancy, poor adaptability to multiple indicators, and insufficient model accuracy, especially in multi-channel spectral detection where it is difficult to achieve high-precision multi-indicator detection.

Method used

By employing a method of characteristic wavelength optimization and multi-model fusion, and through the acquisition of near-infrared diffuse reflectance spectral data across the entire band, combined with random partitioning and spatially balanced random partitioning methods, appropriate preprocessing and dimensionality reduction algorithms are selected to establish partial least squares regression and extreme learning machine models, determine the core light source wavelength, and achieve high-precision detection of multiple indicators in rapeseed.

Benefits of technology

It enables rapid and non-destructive testing of rapeseed quality, reduces hardware costs, simplifies operation, avoids the use of chemical reagents and environmental pollution, meets the needs of large-scale real-time testing in the field, and achieves practical accuracy in predicting various physicochemical indicators.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a rapeseed quality detection method based on characteristic wavelength optimization and multi-model fusion, which comprises the following steps: collecting full-wave band near-infrared diffuse reflection spectrum data of rapeseed samples, synchronously measuring true values of physicochemical indexes of the rapeseed samples, and constructing an original data set; dividing the original data set into a training set and a test set according to the data distribution characteristics of the physicochemical indexes; respectively selecting a preprocessing algorithm to process the full-band near-infrared diffuse reflection spectrum data according to each physicochemical index; aiming at each physicochemical index, respectively selecting a dimension reduction algorithm to extract a characteristic wavelength from the preprocessed spectral data; establishing a prediction model of each physicochemical index based on the training set, verifying the performance of the prediction model by using the test set, and determining an optimal algorithm combination of each physicochemical index; and calculating the comprehensive contribution score of each wavelength point in the full-band spectrum to the four physicochemical indexes by using a variable importance projection algorithm, setting a multi-index comprehensive threshold value, and determining the core light source wavelength of the portable device from a high-score region.
Need to check novelty before this filing date? Find Prior Art