Rapid detection method for pork freshness using fusion spectroscopy and improved BAS-LSSVM

ObjectiveTo realize accurate, rapid, and non-destructive testing of meat freshness.MethodsExtracting spectral feature information based on a spectral acquisition system, proposed a fast non-destructive detection method for meat freshness (TVB-N) by combining an improved beetle whisker search algorithm with least squares support vector machine. By combining SG smoothing filtering and standard normal variables for data preprocessing, combining window competitive adaptive reweighted sampling and iterative continuous projection for feature selection, regularization parameters and kernel parameters of Least-Square Support Vector Machine were optimized by the Improved Beetle Antennae Search Algorithm, a fast non-destructive detection method for meat freshness (TVB-N) was completed. Analyze the performance of the proposed method through experiments.ResultsThe experimental method could achieve accurate, rapid, and non-destructive testing of pork freshness (TVB-N), with high detection accuracy and efficiency, the detection correlation coefficient was 0.978 1, the mean square error was 0.302 1, and the average detection time was 0.031 seconds.ConclusionA fast non-destructive testing method for meat freshness (TVB-N) can be achieved by combining spectral detection and intelligent algorithms.