拉曼光谱结合模拟退火的小麦粉灰分含量检测

Detection of Ash Content of Wheat Flour Based on Raman Spectroscopy Combined with Simulated Annealing

针对当前小麦粉中灰分的检测方法测量周期长,且手工操作复杂,容易产生误差的问题,研究提出采用拉曼光谱技术,结合5种光谱预处理方法和模拟退火算法进行波数筛选两方面优化小麦粉中灰分的拉曼光谱模型。实验结果中,卷积平滑SG(17)+标准正态变换(SNV)+模拟退火(SAA)的优化方式效果最优,相关系数(R2)为0.9875,均方根误差(RMSEC)和预测均方根误差(RMSEP)分别为0.0161和0.15,相对分析误差(RPD)高达8.1679,模型稳健性参数良好。研究结果表明,在模型待测组分浓度与波数相关性以及模型预测准确性方面展现了自身的优越性,该方法简单实用,快速准确,有望将此技术扩展应用于整个食品行业。

Aiming at the current detection method of ash content in wheat flour, the measurement cycle was long, and the manual operation was complex, and it was easy to produce errors. In this paper, Raman spectroscopy was proposed to optimize the Raman spectral model of wheat flour ash by combining five spectral pretreatment methods and simulated annealing algorithm for wave number screening. According to the experimental results, the optimal method of SG (17)+ SNV +SAA was the best, R 2 was 0.987 5, RMSEC and RMSEP were 0.016 1 and 0.15 respectively. And the RPD was as high as 8.167 9. The robustness parameters of the model were good. The results show that the model has demonstrated its own superiority in the correlation between concentration and wave number components of the model to be measured and model prediction accuracy. This method was simple, practical, fast and accurate, and it was expected to extend this technology to the entire food industry.