Conjunctive use of surface and groundwater resources in irrigation networks has a valuable role in the management and planning of water resources with the aim of saving groundwater consumption and increasing the water productivity index. In recent years, linked simulation-optimization models have been recognized as one of the successful solutions to conjunctive use problems. In this paper, the gravitational search meta-heuristic algorithm was used to develop the optimization model and the neural network simulation model was used to estimate the groundwater drawdown. The study area was west of Qazvin plain irrigation network in Iran. In this regard, first, the drought index of the statistical period on the average rainfall of the area was determined by SPI method and three water years consisting o dry, normal and wet in this period was determined. Then, based on the data analysis of piezometers in the study area, this area was divided into two zones: low drawdown or zone 1 ( annual average cumulative drawdown of less than 2.2 meters) and high drawdown zone or zone 2 ( annual average cumulative drawdown of more than 2.2 meters) and 9 scenarios were defined for each area. These scenarios include three crop patterns in dry, wet and normal years. The selected crop patterns included one existing crop pattern and two other patterns based on the crop pattern proposed by Jihad-e-Keshavarzi to the network management. From combining three years of climate index with these three crop patterns in each zone, 9 scenarios and a totally 18 scenarios were determined. The monthly water demand of crops in each zone was calculated each year by calculating the monthly evapotrairation of crops using the modified FAO Penman-Monteith method and using CROPWAT 8 software and finally subtraction from the effective monthly rainfall. Next, the simulation model of each zone was made using an Artificial Neural Network(ANN); This model has 17 series of input data including meteorological, hydrological and water resources data and 5 neurons in the hidden layer and a series of output data including groundwater level variations and was defined as 1-5-17. The optimization model was programmed by the gravitational search algorithm method for each scenario in MATLAB environment. The variables including cultivated areas for each crop, monthly extractions of surface water resources and monthly groundwater extractions of groundwater resource were defined. Eventually the two models were combined. The objective of these combined models is reaching to maximum net economic benefit with the constraint of allowable of drawdown in groundwater level by deficit irrigation and modifying the crop pattern. Based on net economical benefit and water productivity, in all model scenarios except normal year scenarios in zone 1, the best crop patterns are II, I and then existing crop pattern, respectively. In normal year scenarios for zone 1, the best net benefit and the best water productivity is assigned to I, II and then existing crop pattern, respectively. In all scenarios, in different climatic conditions, wet, normal and dry have the highest net economic benefits, respectively. The percentage of total water supply in the scenarios of the zone 1 was estimated between 78 to 93% and in the scenarios of the zone 2 between 76 to 97%. From the view of water supply adequacy in different months, the highest monthly stress occurs in March to June. The highest cultivated areas were related to the first crop pattern scenarios in both zones. The lowest cultivated areas occurred in dry year scenarios. The results show that the simulation-optimization models can decrease the groundwater drawdown in all scenarios by 0.77 m to 1.84 m in zone 1, and by 1.28 m to 1.97 m in zone 2. Furthermore, by replacing the existing crop pattern with two other ones, NB is increased by 35.5% to 53.7% in the zone 1 and by 24.9% to 59.7% in the zone 2. Additionally, the net benefit per unit water consumption volume is increased by 18.9% to 32.2% in zone 1 and by 9% to 52.6% in zone 2. Key Words: Gravitational Search Algorithm (GSA), Artificial Neural Network(ANN), Optimization, Crop pattern, Deficit irrigation, Conjunctive use.