The aim of this study is to determine the amount of consumable energy and its index, the amount of greenhouse gas emissions resulting from energy consumption and choose the best model energy output and green house gas emissions using the model with neural network in dairy farms in Qazvin city of Iran. The results of this study showed; average energy input in dairy farms is 147659.442 MJ per head of cow. Instead this amount of energy input, 23642.25 MJ of energy been produced per head of cow that 91% of it has been related to the milk. An analysis of greenhouse gas emissions showed that for each lactation period, 5393.492kg of greenhouse gas emissions per head of a cow to be released in one year that methane from enteric fermentation has most roles in greenhouse gas emissions in dairy farms. In this study, multi-layer neural networks based on the back propagation algorithm and sigmoid learning function was used for training artificial neural network based on data collected from dairy farms. The artificial neural network model with (5-17-2) structure was the best model for predicting the amount of energy output (milk) and the amount of greenhouse gas emissions. In the best topology, the (R) was calculated as 0.999 and 0.968 for train and test, respectively.
