Prediction of short term traffic speed using LSTM algorithm deep learning

Traffic flow prediction has long been considered as a fundamental problem in an intelligent transportation system. Accurate and timely prediction of the traffic flow situation can be useful for traffic management organizations and individual drivers. A good traffic forecast may help travelers make better travel decisions, thus reducing terrible traffic congestion in cities and carbon dioxide emissions, and improving the efficiency of traffic operations. Recently, various deep learning models have been introduced to the field of prediction. While many studies performed traffic speed prediction with deep learning models. Considering the important effects of spatio-temporal factors and the excellent performance of recurrent neural networks in the field of time series analysis, in the research carried out, the characteristics of long-short-term memory (FI-LSTM) neural networks that combine sequential time data are suggested to be one It is a deep learning network. This research has compared four deep learning models including LSTM, GRU, ConvLSTM, and BiLSTM in three time steps of 5 minutes, 10 minutes, and 15 minutes. The results of this research show that the proposed model compares four deep learning models, as the interval of time steps increases, the errors more and the accuracy of the model decreases, according to the obtained results, the accuracy of the FI-LSTM model is 0.41% higher than the lowest accuracy of the deep learning model (BiLSTM), and also the proposed model for predicting traffic speed in a time step of 5 minutes is 1.34% It has performed better than the time step of 10 minutes.