基于GA和CARS的真空包装冷却羊肉细菌菌落总数高光谱检测

发布时间：2018-08-11 09:45

【摘要】：在光谱建模过程中,采用不同的变量筛选算法进行光谱特征波段的提取已成为提高模型效果的重要方法。以真空包装的冷却羊肉细菌菌落总数作为研究指标,比较了两种变量筛选算法对其高光谱偏最小二乘(partial least squares,PLS)模型效果的影响。研究提取了样品肌肉感兴趣区域(ROIs)的羊肉光谱并进行预处理,进而采用遗传算法(genetic algorithm,GA)和竞争性自适应重加权法(competitive adaptive reweighted sampling,CARS)分别对预处理后的473~1 000 nm范围光谱进行特征波段的提取,对比分析了不同波段下羊肉细菌菌落总数的GA-PLS,CARS-PLS和全波段PLS(W-PLS)模型效果。结果表明,GA-PLS和CARS-PLS的模型效果均优于W-PLS,且CARS-PLS模型效果最好,其校正集的决定系数(R_c~2)和均方根误差(root mean square error,RMSEC)分别为0.96和0.29,交互验证的决定系数(R_(cv)~2)和均方根误差(root mean square errorof cross validation,RMSECV)分别为0.92和0.46,预测集的决定系数(R_p~2)和均方根误差(root mean square error of prediction,RMSEP)分别为0.92和0.47,预测相对分析误差(relative prediction deviation,RPD)为3.58。因此利用高光谱图像技术结合CARS-PLS可以实现羊肉细菌菌落总数快速无损准确检测。
[Abstract]:In the process of spectral modeling, it has become an important method to improve the effectiveness of the model by using different variable selection algorithms to extract spectral feature bands. The effects of two variable selection algorithms on the hyperspectral partial least squares (partial least squares) model were compared with the total bacterial colony count of chilled mutton in vacuum packaging. The mutton spectra of the muscle region of interest (ROIs) were extracted and pretreated, and the characteristic bands were extracted by genetic algorithm (GA) and competitive adaptive reweighting method (competitive adaptive reweighted sampling car), respectively. The effects of GA-PLS CARS-PLS model and full-band PLS (W-PLS) model of mutton bacterial colony in different bands were compared and analyzed. The results show that the model effect of GA-PLS and CARS-PLS is better than that of W-PLS, and the effect of CARS-PLS model is the best. The determination coefficient (R_c~2) and root mean square error (root mean square error) of the corrected set are 0.96 and 0.29, respectively. The determination coefficients (R _ (cv) _ 2) and the root mean square error (root mean square errorof cross) of the calibration set are 0.92 and 0.46, respectively. The determination coefficient (R_p~2) and the root mean square error (root) of the predicted set are 0.92 and 0.46, respectively. Mean square error of prediction RMSEP was 0. 92 and 0. 47, respectively. The relative analysis error (relative prediction deviation RPD) was 3. 58. Therefore, the use of hyperspectral image technology combined with CARS-PLS can achieve rapid and accurate detection of mutton bacterial colonies.
【作者单位】：石河子大学机械电气工程学院;石河子大学食品学院;
【基金】：国家自然科学基金项目(31460418) 高等学校博士学科点专项科研基金项目(2013651820004)资助
【分类号】：TS251.7;O657.3

【相似文献】