Application of Artificial Neural Networks for Predicting the Yield and GHG Emissions of Sugarcane Production
Introduction One of the most important sources of the sugar production is sugarcane.Sugar is one of the eight human food sources (wheat, rice, corn, sugar, cattle, sorghum, millet and cassava). Also sugarcane is mainly used for livestock feed, electricity generation, fiber and fertilizer and in many...
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN FA |
| Publicado: |
Ferdowsi University of Mashhad
2018
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/c0ce5b1191904131815b1a839f887e31 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Introduction One of the most important sources of the sugar production is sugarcane.Sugar is one of the eight human food sources (wheat, rice, corn, sugar, cattle, sorghum, millet and cassava). Also sugarcane is mainly used for livestock feed, electricity generation, fiber and fertilizer and in many countries sugarcane is a renewable source for the biofuel. The efficient use of inputs in agriculture lead to the sustainable production and help to reduce the fossil fuel consumption and greenhouse gases emission and save financial resources. Furthermore, detecting relationship between the energy consumption and the yield is necessary to approach the sustainable agriculture. It is generally accepted that many countries try to reduce their dependence to agricultural crop productions of other countries. The being Independent on agricultural productions lead to take more attention to modern methods and the objective of all these methods is increasing the performance with the efficient use of inputs or optimizing energy consumptions in agricultural systems. Energy modeling is a modern method for farm management that this model can predict yield with using the different amount of inputs. The objective of this study was to predict sugarcane production yield and (greenhouse gas) GHG emissions on the basis of energy inputs. Materials and Methods This study was carried out in Khouzestan province of Iran. Data were collected from 55 plant farms in Debel khazai Agro-Industry using face to face questionnaire method. In this study, the energy used in the sugarcane production has considered for the energy analysis without taking into account the environmental sources of the energy such as radiation, wind, rain, etc. Energy consumption in sugarcane production was calculated based on direct and indirect energy sources including human, diesel fuel, chemical fertilizers, pesticides, machinery, irrigation water, electricity and sugarcane stalk. Energy values were calculated by multiplying inputs and outputs per hectare by their coefficients of energy equivalents. Input energy in agricultural systems includes both direct and indirect energy and renewable and non-renewable forms. Direct energies include human labor, diesel fuel, water for irrigation and electricity and indirect energies consisted of machinery, seed (cultivation of sugarcane has been done with cutting of sugarcane instead of seed), chemical fertilizer. Renewable energies include machinery, sugarcane stalk, chemical fertilizer while non-renewable energy consisted of machinery, chemical fertilizer, electricity and diesel fuel. Energy values were calculated by multiplying inputs and outputs per hectare by their coefficients of energy equivalents. The amounts of GHG emissions from inputs in sugarcane production per hectare were calculated by CO2 emissions coefficient of agricultural inputs. Energy modeling is an attractive subject for engineers and scientists who are concerned about the energy management. In the energy area, many different of models have been applied for modeling future energy. An artificial neural network (ANN) is an artificial intelligence that it can applied as a predictive tool for nonlinear multi parametric. Artificial neural network has been applied successfully in structural engineering modeling ANNs are inspired by biological neural networks. Results and Discussion The total energy used in the farm operations during the sugarcane production and the energy output was 1742883.769 and 111000 MJha_1, respectively. Electricity (52%) and chemical fertilizers (16%) were the most influential factors in the energy consumption. The electricity contribution was the highest due to the low efficiency of energy conversion in electric motors which were used for irrigation in the study area. In some areas, inefficient surface irrigation wastes a lot of water and energy (in forms of electricity). Another reason is that electricity energy equivalent for Iranian electricity production is higher than developed countries because Iran’s electricity grid is highly dependent on fossil fuels, so that 95% of the electrical energy in Iran is generated in thermal power plants using fossil fuels sources. In addition, the electricity transmission system is too old. GHG emissions data analysis indicated that the total GHG emissions was 415337.62 kg ha-1 (CO2eq) kgCO2eq ha-1 in which burning trash with the share of 62% had the highest GHG emission and followed by electricity (32%), respectively. The ANN model with 7-5-15-1 and 5-5-1 structure were the best model for predicting the sugarcane yield and GHG emissions, respectively. The coefficients of determination (R2) of the best topology were 0.98 and 0.99 for the sugarcane yield and GHG emissions, respectively. The values of RMSE for sugarcane production and GHG emission were found to be 0.0037 and 4.52×10-6, respectively. Conclusions The statistical parameters of R2 and RMSE demonstrated that the proposed artificial neural networks results have best accuracy and can predict the yield and GHG emission. It is generally showed that artificial neural networks have good potential to predict the yield of the sugarcane production. |
|---|