A Hybrid Deep Learning Method for Short-Term Traffic Flow Forecasting: GSA-LSTM
Bharti Naheliya, Poonam Redhu, Kranti Kumar
Abstract
Objectives: The main objective of this study is to improve the accuracy and reliability of the short-term traffic flow forecasting method, while simultaneously addressing limitations in existing models and proposing a novel approach that enhances the quality of the traffic flow predictions. Methods: This study developed a long short-term memory (LSTM) neural network optimized by the gravitational search approach (GSA) to enhance prediction accuracy for short-term traffic flow. The gravitational search algorithm selects the best parameters for the long short-term memory neural network on a global scale. Findings: The proposed GSA-LSTM model exhibits significant superiority over the selected models when assessing performance through the evaluation metrics such as root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and correlation coefficient (r). Moreover, the average accuracy of the proposed model is higher as compared to other existing neural network models which depicted the effectiveness of the proposed model. Novelty: Tables and figures displayed that the performance accuracy of the proposed model is higher than the other selected models such as autoregressive integrated moving average (ARIMA), wavelet neural network (WNN), Gated Recurrent Unit (GRU), long short-term memory model (LSTM), GSA-GRU, ACO-LSTM, and PSO-LSTM model. Keywords: Intelligent Transportation System, Deep Learning, Traffic Flow Prediction