Energy audit of Iranian kiwifruit production using intelligent systems

Abstract

Optimizing the energy flows of agricultural production is a concern in order to find the most appropriate mix of agricultural inputs, which would in turn minimize energy consumption and maximize energy output. Thus, the aim of this study is to model the energy flows of kiwifruit production in Guilan province of Iran (as a case study) using Artificial Neural Network (ANN) + Genetic Algorithm (GA) modeling and Multiple Linear Regressions (MLR) + GA modeling. The results indicated that the highest energy consumption were attributed to electricity and chemical fertilizers with the shares of 42% and 25%, respectively. Energy indices such as energy use efficiency, energy productivity, specific energy and net energy were determined to be 0.48, 0.25 kgMJ−1, 4.01 MJkg−1, and -54,644 MJha−1. The performance indices such as coefficient of determination (R2) and efficiency (EF) for the best MLR model were determined to be 0.61 and 0.60%, respectively. Moreover, the same indices for the best developed ANN model were 0.73 and 0.72%, respectively. Overall, it was concluded that the ANNs models could better predict the energy output than the MLRs models and the performance of ANN highlighted that this model may be applied to prognosticate the energy output of kiwifruit production. To conclude, a comparison between ANN + GA and MLR + GA clearly demonstrated the better performance of ANN + GA to optimize the energy flows of Iranian kiwifruit production.