Combination of hyperspectral sensing images and chemometrics for measuring tensile strength indices of organic plastic sheeting in the field

ABSTRACT

In this paper, hyperspectral sensing image and chemometrics were combined to measure the tensile strength of the organic plastic sheeting of ethylene-vinyl acetate copolymer (EVAC). Three different brands of EVAC sheeting materials were used in the investigation. Spectral data of 978–1676 nm agricultural greenhouse EVAC sheeting were collected by near infrared hyperspectral imager in closed black box. The tensile strength analysis device is used to evaluate the tensile strength indices. The appropriate chemometrics is adopted to build the associated model between the feature spectral data, which expresses the cross-linking attributes of the inner organic molecules, and the tensile strength indices. The cross-validation methods are performed to measure the accuracy and robustness of the training model. The established associated model is then able to evaluate the tensile strength indices of the unknown EVAC sheeting instances. With the leading machine learning algorithm, the corresponding support vector machine regression shows the best detection results, with the percentage of square correlation coefficient ($X$) to be 86.63% and the lower root mean square error ($X$) to be 1.049. It seems that the strategy, which integrates the hyperspectral sensing technology and chemometrics, can be implemented in real-time measuring of the tensile strength indices of EVAC organic plastic sheeting according to the experimental results.

KEYWORDS:

• Infrared
• chemometrics
• sheeting
• tensile strength
• cross-linking

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [5779161].