A portable rapeseed quality non-destructive inspection device based on multichannel spectroscopy

Abstract

It is essential to develop low-cost, rapid and portable systems for detecting the quality of rapeseed planting, harvesting, and storage. A multichannel spectral detection system for the quantitative assessment of rapeseed oil, protein, glucosinolate, and moisture content was developed in this study. The core hardware of the system comprises a custom-designed spectral acquisition module and a Raspberry Pi. The spectral module consists of a spectrum sensor and a characteristic wavelength LED, featuring 10 channels with a wavelength range of 850–1550 nm included. The results from the test set indicate that the most accurate oil predictions can be achieved using the SPXY+SNV+CARS+PLS method. For protein predictions, the optimal results were obtained using the Random+MSC +UVE+PLS approach. The best predictions for glucosinolates and moisture content were achieved with the Random+SNV+CARS+PLS method. To verify the performance of this systems, independent data were used for external validation. The RMSE, R², MAE results for oil, protein, glucosinolates, and moisture were 2.04 %, 0.69, 1.58 %, 1.52 %, 0.67, 1.25 %, 18.86μmol·g⁻¹, 0.52, 15.03μmol·g⁻¹, 0.36 %, 0.74, 0.33 %, respectively. In general, the developed detection system has potential for rapid detection of rapeseed in the field or market.

Code and data availability

公開状態または取得可能な本文経路を確認できませんでした。

No evidence-backed public reproduction asset is currently recorded.