Department of Agricultural Machinery Engineering, Faculty of Agricultural Engineering and Technology, University of Tehran, Karaj, Iran.
Department of Agricultural Machinery Engineering, Faculty of Agricultural Engineering and Technology, University of Tehran, Karaj, Iran.
Sci Total Environ. 2019 May 10;664:1005-1019. doi: 10.1016/j.scitotenv.2019.02.004. Epub 2019 Feb 6.
This study aims to employ two artificial intelligence (AI) methods, namely, artificial neural networks (ANNs) and adaptive neuro fuzzy inference system (ANFIS) model, for predicting life cycle environmental impacts and output energy of sugarcane production in planted or ratoon farms. The study is performed in Imam Khomeini Sugarcane Agro-Industrial Company (IKSAIC) in Khuzestan province of Iran. Based on the cradle to grave approach, life cycle assessment (LCA) is employed to evaluate environmental impacts and study environmental impact categories of sugarcane production. Results of this study show that the consumed and output energies of sugarcane production are in average 172,856.14 MJ ha, 120,000 MJ ha in planted farms and 122,801.15 MJ ha, 98,850 MJ ha in ratoon farms, respectively. Results show that, in sugarcane production, electricity, machinery, biocides and sugarcane stem cuttings have the largest impact on the indices in planted farms. However, in ratoon farms, electricity, machinery, biocides and nitrogen fertilizers have the largest share in increasing the indices. ANN model with 9-10-5-11 and 7-9-6-11 structures are the best topologies for predicting environmental impacts and output energy of sugarcane production in planted and ratoon farms, respectively. Results from ANN models indicated that the coefficient of determination (R) varies from 0.923 to 0.986 in planted farms and 0.942 to 0.982 in ratoon farms in training stage for environmental impacts and outpt energy. Results from ANFIS model, which is developed based on a hybrid learning algorithm, showed that, for prediction of environmental impacts, R varies from 0.912 to 0.978 and 0.986 to 0.999 in plant and ratoon farms, respectively, and for prediction of output energy, R varies from 0.944 and 0.996 in planted and ratoon farms. Results indicate that ANFIS model is a useful tool for prediction of environmental impacts and output energy of sugarcane production in planted and ratoon farms.
本研究旨在采用两种人工智能（AI）方法，即人工神经网络（ANNs）和自适应神经模糊推理系统（ANFIS）模型，预测种植或宿根蔗田甘蔗生产的生命周期环境影响和输出能量。该研究在伊朗胡齐斯坦省的伊玛目霍梅尼甘蔗农业综合企业（IKSAIC）进行。基于摇篮到坟墓的方法，生命周期评估（LCA）用于评估环境影响并研究甘蔗生产的环境影响类别。本研究的结果表明，甘蔗生产的消耗和输出能量平均为 172856.14 MJ ha，种植农场为 120000 MJ ha，宿根农场为 122801.15 MJ ha，98850 MJ ha。结果表明，在甘蔗生产中，电力、机械、杀菌剂和甘蔗茎段切割对种植农场的指数有最大的影响。然而，在宿根农场中，电力、机械、杀菌剂和氮肥在增加指数方面的份额最大。具有 9-10-5-11 和 7-9-6-11 结构的 ANN 模型分别是预测种植和宿根蔗田甘蔗生产环境影响和输出能量的最佳拓扑结构。ANN 模型的结果表明，在种植农场的训练阶段，环境影响和输出能量的决定系数（R）从 0.923 到 0.986 不等，在宿根农场的训练阶段，R 从 0.942 到 0.982 不等。基于混合学习算法开发的 ANFIS 模型的结果表明，对于环境影响的预测，R 在种植和宿根农场中分别从 0.912 到 0.978 和 0.986 到 0.999 不等，对于输出能量的预测，R 在种植和宿根农场中分别从 0.944 和 0.996 不等。结果表明，ANFIS 模型是预测种植和宿根蔗田甘蔗生产环境影响和输出能量的有用工具。
