Short-term traffic flow prediction affected by climatic conditions based on neural networks approach

Because economic expansion, industrial growth, and urban population growth have increased traffic congestion, reliable traffic flow forecasts are essential for traffic monitoring and management. However, time-series techniques may neglect geographical and transitory meteorological factors in short-term traffic flow estimates. This research introduces a hybrid deep learning model that combines One-Dimensional Convolutional Neural Networks (۱DCNN) with Long Short-Term Memory (LSTM) to anticipate traffic flow. The model for the Neyshabur-Mashhad axis includes meteorological factors to enhance forecasts. One-dimensional interpolation extracts spatial information from traffic data; LSTM captures temporal correlations. Experimental results demonstrate that the ۱DCNN-LSTM model outperforms prior models, especially in weather. While AdaGrad performs poorly on large datasets, Adam optimization improves prediction accuracy. These findings demonstrate the potential of the intelligent transportation system model, giving insights for infrastructure construction and real-time traffic management. This study uses topographical and climatic characteristics to reduce traffic congestion, improve road safety, and increase urban mobility.

کلیدواژه ها:

Traffic flow prediction ، convolutional neural network ، Long Short-Term Memory ، Deep Learning ، Weather