Energy Consumption and Modeling of output energy with Multilayer Feed-Forward Neural Network for Corn Silage in Iran

Abstract

In this study, various Artificial Neural Networks (ANNs) were developed to estimate the output energy for corn silage production in Esfahan province, Iran. For this purpose, the data on 65 corn silage production farms in the Esfahan province, were collected and analyzed. The results indicated that total energy input for corn silage production was about 83126 MJ ha^–1; machinery (with 38.8 %) and chemical fertilizer (with 24.5 %) were amongst the highest energy inputs for corn silage production. The developed ANN was a multilayer perceptron (MLP) with eight neurons in the input layer (human power, machinery, diesel fuel, chemical fertilizer, water for irrigation, seed, farm manure and pesticides ), one, two, three, four and five hidden layer(s) of various numbers of neurons and one neuron (output energy) in the output layer. The results of ANNs analyze showed that the (8-5-5-1)-MLP, namely, a network having five neurons in the first and second hidden layer was the best-suited model estimating the corn silage output energy. For this topology, MAB, MAE, RMSE and R² were 0.109, 0.001, 0.0464 and 98%, respectively. The sensitivity analysis of input parameters on output showed that diesel fuel and seeds had the highest and lowest sensitivity on output energy with 0.0984 and 0.0386, respectively. The ANN approach appears to be a suitable method for modeling output energy, fuel consumption, CO₂ emission, yield, and energy consumption based on social and technical parameters. This method would open new doors to advances in agriculture and modeling.