基于信息融合的小麦粉品质快速检测

Rapid determination of wheat flour quality based on data fusion technology

为探索信息融合技术对小麦粉品质快速检测的可行性,试验采集小麦粉的近红外光谱及中红外光谱,建立基于信息融合的小麦粉品质指标(蛋白质含量、湿面筋含量、吸水量、形成时间、稳定时间、弱化度)快速检测模型,并采用前向区间变量筛选及遗传算法对信息融合模型进行了优化。研究表明,使用前向区间算法进行变量筛选后,信息融合模型的预测能力大幅度提升;再继续使用遗传算法优化光谱变量可以进一步提高信息融合模型的预测能力,并简化模型所用的变量数量。试验构建的最优信息融合模型对蛋白质含量、湿面筋含量、吸水量、形成时间、稳定时间和弱化度的预测相关系数(r)分别达到了0. 98、0. 98、0. 97、0. 94、0. 95和0. 95,预测误差均方根(root mean square error of prediction,RMSEP)分别为0. 181、0. 590、0. 455、0. 502、0. 557和13. 047。试验结果表明,采用信息融合技术对小麦粉品质进行快速检测是可行的。

The aim of the study was to explore the feasibility of applying information fusion technology in rapid detection of wheat flour quality. Near-infrared and mid-infrared spectra of wheat flour were collected, and a fast predicting model for wheat flour quality indices (protein content, wet gluten content, water absorption amount, dough development time, dough stability, and degree of softening) based on data fusion was established, followed by forward interval variable selection algorithm and genetic algorithm optimization. The results showed that the prediction correlation coefficients ( r ) of the optimal data fusion model for protein content, wet gluten content, water absorption amount, dough development time, dough stability, and degree of softening were 0.98, 0.98, 0.97, 0.94, 0.95 and 0.95, respectively, and the root mean square error of prediction (RMSEP) were 0.181, 0.590, 0.455, 0.502, 0.557 and13.047, respectively. In conclusion, it is feasible to rapidly predict wheat flour quality by using data fusion technology.