基于手持式近红外光谱仪的三文鱼菌落总数 检测技术

Detection of total number of salmon colonies by handheld near infrared spectrometer

中文关键词:  三文鱼  近红外光谱  菌落总数  BP神经网络

英文关键词:salmon  near infrared spectroscopy  total numbers of colony  back-propagation artificial neural network

中文摘要:

      目的 通过对近红外广谱数据进行神经网络系统训练, 讨论近红外广谱技术对冷藏三文鱼菌落总数快速预测的可行性。方法 针对三文鱼在4 ℃贮藏过程中的微生物变化, 利用手持式近红外光谱仪, 通过小波分析对于光谱进行预处理, 之后结合遗传算法和BP神经网络系统方法建立预测和检测模型。结果 该模型与传统平板计数方法的相关系数为0.981, 均方根误差为0.097, 验证模型的相关系数为0.960, 均方根误差为0.098, 具有良好精确度、准确度。结论 该方法能够用于冷藏三文鱼菌落总数的无损、现场检测。

英文摘要:

      Objective To develop a new method by using artificial neural network for discussing the feasibility of predicting the aerobic plate count of salmon. Methods After spectral pretreatment by wavelet analysis, a new prediction and validation model was established by using a combined tactic of genetic algorithm (GA) and back-propagation artificial neural network (BP-ANN) to predict the aerobic palate count of salmon based on the change of microbe during the storage at 4 ℃, and portable near infrared spectrometer was used. Results The model had high accuracy and precision, the calibration curve coefficient of correlation (R) of the model and the traditional plate count method was 0.981, and root mean square error (RMSE) was 0.097. Correlation coefficient of validation model was 0.960 and root mean square error (RMSE) was 0.098. Conclusion This model could be used for non-destructive and on-site detection of the total bacteria colonies in frozen salmon.