Razavi Mahsa Sadat, Golmohammadi Abdollah, Sedghi Reza, Asghari Ali
Mechanical Engineering of Biosystems Department of Biosystems Engineering Faculty of Agricultural and Natural Resources University of Mohaghegh Ardabili Ardabil Iran.
Department of Biosystems Engineering Faculty of Agricultural and Natural Resources University of Mohaghegh Ardabili Ardabil Iran.
Food Sci Nutr. 2019 Dec 26;8(2):884-893. doi: 10.1002/fsn3.1365. eCollection 2020 Feb.
Bruises occur under both static and dynamic loadings when the imposed stress on fruit goes over the failure stress of the fruit tissue. Bruise damage is the main reason for fruit quality loss. In this study, the potential of artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), and multiple regression (MR) techniques to predict bruise volume propagation of pears during the storage time was evaluated. For this purpose, at first, the radius of curvature at loading region was obtained. Samples were divided into five groups and subjected to five force levels. Then, they were kept under storage conditions and at 7-time intervals after loading tests, bruise volume was calculated using magnetic resonance imaging (MRI) and image processing techniques. Force, storage time, and radius of curvature at loading region were employed as input variables, and bruise volume (BV) was considered as output in the developed models. Multilayer perceptron (MLP) artificial neural network with three layers that includes an input layer (three neurons), two hidden layers (two and nine neurons), and one output layer was used. For the evaluation of models, three criteria (RMSE, VAF, and ) were calculated. ANN and MR gave the highest and lowest correlation between predicted and actual values, respectively. These results indicate that the ANN techniques can be used to predict pear bruising propagation in storage time.
当施加在水果上的应力超过水果组织的破坏应力时，无论是在静态还是动态载荷下都会出现瘀伤。瘀伤损害是水果品质下降的主要原因。在本研究中，评估了人工神经网络（ANN）、自适应神经模糊推理系统（ANFIS）和多元回归（MR）技术预测梨在储存期间瘀伤体积扩展的潜力。为此，首先获取加载区域的曲率半径。将样本分为五组，并施加五个力水平。然后，将它们置于储存条件下，并在加载测试后的7个时间间隔，使用磁共振成像（MRI）和图像处理技术计算瘀伤体积。将力、储存时间和加载区域的曲率半径用作输入变量，在开发的模型中，将瘀伤体积（BV）视为输出。使用了具有三层的多层感知器（MLP）人工神经网络，包括一个输入层（三个神经元）、两个隐藏层（两个和九个神经元）和一个输出层。为了评估模型，计算了三个标准（RMSE、VAF和 ）。ANN和MR分别在预测值和实际值之间给出了最高和最低的相关性。这些结果表明，ANN技术可用于预测梨在储存期间的瘀伤扩展情况。
