Qiao Mengmeng, Xia Guoyi, Xu Yang, Cui Tao, Fan Chenlong, Li Yibo, Han Shaoyun, Qian Jun
College of Engineering, China Agricultural University, Beijing, People's Republic of China.
Universität Bremen, Bremen, Germany.
J Sci Food Agric. 2024 Aug 30;104(11):6594-6604. doi: 10.1002/jsfa.13483. Epub 2024 Apr 3.
The rapid and accurate detection of moisture content is important to ensure maize quality. However, existing technologies for rapidly detecting moisture content often suffer from the use of costly equipment, stringent environmental requirements, or limited accuracy. This study proposes a simple and effective method for detecting the moisture content of single maize kernels based on viscoelastic properties.
Two types of viscoelastic experiments were conducted involving three different parameters: relaxation tests (initial loads: 60, 80, 100 N) and frequency-sweep tests (frequencies: 0.6, 0.8, 1 Hz). These experiments generated corresponding force-time graphs and viscoelastic parameters were extracted based on the four-element Maxwell model. Then, viscoelastic parameters and data of force-time graphs were employed as input variables to explore the relationships with moisture content separately. The impact of different preprocessing methods and feature time variables on model accuracy was explored based on force-time graphs. The results indicate that models utilizing the force-time data were more accurate than those utilizing viscoelastic parameters. The best model was established by partial least squares regression based on S-G smoothing data from relaxation tests conducted with initial force of 100 N. The correlation coefficient and the root mean square error of the calibration set were 0.954 and 0.021, respectively. The corresponding values of the prediction set were 0.905 and 0.029, respectively.
This study confirms the potential for accurate and fast detection of moisture content in single maize kernels using viscoelastic properties, which provides a novel approach for the detection of various components in cereals. © 2024 Society of Chemical Industry.
快速准确地检测水分含量对于确保玉米质量至关重要。然而，现有的快速检测水分含量的技术常常存在设备成本高、环境要求苛刻或准确性有限等问题。本研究提出了一种基于粘弹性特性检测单个玉米粒水分含量的简单有效方法。
进行了两种类型的粘弹性实验，涉及三个不同参数：松弛试验（初始载荷：60、80、100 N）和频率扫描试验（频率：0.6、0.8、1 Hz）。这些实验生成了相应的力-时间图，并基于四元件麦克斯韦模型提取了粘弹性参数。然后，将粘弹性参数和力-时间图的数据作为输入变量，分别探讨与水分含量的关系。基于力-时间图，研究了不同预处理方法和特征时间变量对模型准确性的影响。结果表明，利用力-时间数据的模型比利用粘弹性参数的模型更准确。基于100 N初始力进行的松弛试验的S-G平滑数据，通过偏最小二乘回归建立了最佳模型。校准集的相关系数和均方根误差分别为0.954和0.021。预测集的相应值分别为0.905和0.029。
本研究证实了利用粘弹性特性准确快速检测单个玉米粒水分含量的潜力，为谷物中各种成分的检测提供了一种新方法。© 2024化学工业协会。
