摘要
应用便携式近红外仪和近红外光谱分析技术,采用光谱小波变换(WT)的细节系数和偏最小二乘法(PLS),对大黄鱼总脂肪含量建立定标模型。定标集样品55个,内部验证决定系数R2为0.850 4,相对预测标准误差(RSEC)为7.02%,预测集样品24个样,外部验证决定系数R2为0.870 6,相对预测标准误差(RSEP)为7.21%。结果表明,利用小波变换细节系数能有效提取近红外短波区的有用信息,消除光谱背景和噪声;近红外光谱分析技术可以准确、快速、无损害检测大黄鱼的总脂肪含量,利用便携式近红外仪可以实现现场检测。
Based on portable near infrared spectrophotometers,a rapid nondestructive method was developed for predicting the total fat content of Pseudosciaena crocea by near infrared spectroscopy(NIRS).Adopting the detail coefficients of spectra's wavelet transform(WT) and partial least squares (PLS) regression,the prediction model was built on 55 samples.The determination coefficient and RSEC of the model were 0.850 4 and 7.02%,respectively.Applying the model to the test set with 24 samples,the determination coefficient and RSEP were 0.870 6 and 7.21%,respectively.The experimental results show that,we can distill available information,remove fluctuating background and noise from the spectra by detail coefficients of WT;NIRS can be used as a method to detect the total fat content of Pseudosciaena crocea rapidly and nondestructively;in situ determination could be achieved based on a portable near infrared spectrophotometer.
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第S1期59-62,共4页
Periodical of Ocean University of China
基金
国家自然科学基金:水产品中多聚磷酸盐水解酶研究(30671632)
国家高技术研究发展计划项目(2006AA092444)资助
