With the uncontrolled expansion of large cities and the development of industries, air pollution has become a serious issue in urban management. In recent years, repeated droughts and widespread dam construction in neighboring countries have turned the problem of dust particles into a major challenge for Iran and other countries in the region. In many large cities, the concentration of air pollutants exceeds standards, which has widespread negative effects on human health, including an increase in cardiovascular and respiratory diseases, diabetes, hypertension, dementia, miscarriages, and premature deaths. Air pollution also negatively impacts the health of other living organisms, economic and social activities, agriculture, and the environment. The most important air pollutants include suspended particles with a diameter of less than 10 and 2.5 microns, nitrogen dioxide, nitrogen monoxide, sodium sulfate, carbon monoxide, and ozone. In this research, we first examine the issue of air pollution in Lorestan province based on data collected from the air pollution monitoring stations of the Environmental Protection Organization, and we calculate the correlation between the concentration of air pollutants and meteorological variables. Then, we use a multilayer neural network with a stable online learning algorithm to predict short-term pollutant concentration levels. Predicting air pollution is of high importance, and timely predictions can be effective in reducing the negative effects of air pollution. There are various methods for predicting air pollution, some of which have low accuracy. However, the use of artificial neural networks, as a more advanced method, has recently gained attention for modeling and predicting air quality. These models are especially useful for predicting time series and environmental data. Considering the quality of data collected at the air pollution monitoring stations of the Environmental Protection Organization in Lorestan province, the significant impacts of the dust phenomenon, and the geographical locations of the cities of Khorramabad and Poldokhtar, these cities were selected for case studies. According to the obtained results, there is a significant correlation between meteorological variables and pollutant concentrations, and the proposed neural network model demonstrates high accuracy in predicting pollutant concentrations.