Rapid and non-destructive detection of hyperspectral milk protein based on improved WOA-Elman neural network

Abstract

Objective: To solve the problems of low accuracy, low efficiency, and strong manual dependence in existing milk protein detection methods. Methods: Based on hyperspectral imaging systems, proposed a combination of improved whale algorithm and Elman neural network for rapid and non-destructive detection of milk protein content. Optimized the whale algorithm through three aspects (chaotic mapping, adaptive convergence factor, and adaptive weight) to improve search accuracy, and optimized the Elman neural network parameters (weights and thresholds) after optimization. Analyzed the performance of the proposed non-destructive testing method through experimental analysis. Results: Compared with conventional detection methods, proposed method was optimal for multiple performance indicators in non-destructive testing of milk protein. The experimental method was optimal in multiple performance indicators for non-destructive testing of milk protein, with determination coefficient of 0.997 3, the root mean square error of 0.000 3, and the detection time of 1.56 seconds. Conclusion: The experimental method has high detection accuracy and efficiency.

Keywords

• milk
• protein
• non destructive testing
• hyperspectral imaging
• whale algorithm
• elman neural network