Non destructive detection of kiwifruit sugar content based on improved WOA-LSSVM and hyperspectral analysis

Abstract

Objective: Addressing the issues of poor accuracy and low efficiency in non-destructive testing methods for kiwifruit sugar content. Methods: Proposing a non-destructive testing method for kiwifruit sugar content that combined hyperspectral detection technology, least squares support vector machine, and improved whale algorithm. By collecting hyperspectral information of kiwifruit through a hyperspectral detection system, after preprocessing and feature wavelength screening, and then input into an improved whale algorithm optimized least squares support vector machine model to achieve rapid and non-destructive detection of kiwifruit sugar content, and verify its performance. Results: The proposed method could achieve rapid and non-destructive detection of kiwifruit sugar content, with a determination coefficient of 0.965 2 for the test set, a root mean square error of 0.880 5 for the test set, and an average detection time of 1.06 seconds. Conclusion: Combining machine learning algorithms with hyperspectral detection technology can achieve rapid and non-destructive detection of kiwifruit sugar content.

Keywords

• kiwi fruit
• hyperspectral detection
• sugar content
• machine learning algorithms
• whale optimization algorithm
• least squares support vector machine