Dust is a phenomenon that has many destructive environmental effects in different parts of human life, including: agriculture, economics, health and etc. The country of Iran, especially its western and southwestern regions, is suffering a lot of damage due to its presence in the area affected by the dust phenomenon every year. So paying attention to this issue and reducing the resulting damagest is a priority. The purpose of this Research is to investigate and predict the dust phenomenon in southwest of Iran. For this Research, 27 year old Dust Data were used at 14 synoptic Stations in Southwest of Iran during the period (1990-2017). In this Research, dust data was first normalized in 14 stations then, by using the hybrid-panel data model, the ANFIS Compatible Neural Network in Matlab Software was falsified and predicted and finally, to prioritize more stations, dust was exposed to TOPSIS and SAW multivariate decision making models. The findings of the Research showed that the reliability of the lira faction models (neural network of the hybrid panel compared to the ANFIS Comparative neural network) was higher. Based on prediction models, the maximum probability of occurrence, the maximum dust in the next 23 years in the studied area at two stations, Sarpol Zahab and Abadan are respectively (120. 709, 128. 917). According to the SAW model, the probability of occurrence of dust in the next 23 years is estimated at Abadan station with 0. 99% and Based on the TOPSIS model, Islamabad e Gharb station with a value of 97%. In order to reduce the damage caused by the dust phenomenon in the study area, in addition to domestic measures, such as inter-organizational cooperation, it should be addressed by concluding an international agreement with the neighboring countries.