Freeway Short-Term Travel Time Prediction Based on Data Mining

This paper takes the freeway as the research object to explore short-term travel time prediction. Travel time is an important factor in reflecting the traffic condition. Meanwhile, the research of short-term prediction, which needs to be based on big data, is an essential part of intelligent transportation systems. This paper selects the gantry data of Guangdong Shenzhen Freeway to ensure the safety of toll data. It proposes a data-cleaning method that provides a reliable data basis for the prediction. In this paper, the ARIMA, LSTM, and GA-LSTM models are constructed to predict the travel time of the freeway. The conclusion that the GA-LSTM model has the best prediction effect is obtained using the evaluation system composed of MAE and MAPE. This paper finally forecasts the travel time of continuous freeway sections, which is significant to the intelligent transportation system.