|
چکیده
|
Purpose: This study aimed to forecast diesel demand in Iran’s agricultural
sector to support informed energy planning and economic management.
Given diesel fuel's dominant role in agricultural activities, accurate
demand prediction is essential to improve fuel allocation and reduce
inefficiencies in this sector.
Methodology: The present study employed two artificial intelligence–
based approaches, namely artificial neural networks (ANN) and genetic
algorithms (GA), to model and predict diesel demand using historical
annual data from 1994/95 to 2023/24, including diesel consumption,
diesel price, agricultural gross domestic product, mechanization index, and
cultivated area. An ANN model was developed using a feedforward
multilayer perceptron architecture, while a GA was applied as an
independent modeling and optimization approach. Model performance
was evaluated by statistical indicators such as MSE, RMSE, MAE, and
MAPE.
Findings: The results indicated that both models captured the nonlinear
relationships influencing diesel demand in Iran’s agricultural sector.
However, the ANN model showed higher predictive accuracy than the
GA model across all evaluation criteria. Forecast results suggested a
gradual decline in diesel demand over the period 2023/24–2032/33,
reflecting the modeled effects of key economic and structural variables
rather than assumed policy or technological shifts.
Originality/Value: The originality of this research lies in the comparative
application of ANN and GA for forecasting diesel demand in Iran’s
agricultural sector using long-term national data. The findings provide
practical insights for energy planners by identifying a reliable AI-based
forecasting tool tailored to the economic characteristics of agriculture in
Iran.
|