Improved Long Short-Term Memory-Based Periodic Traffic Volume Prediction Method

In response to the problem of fixed time intervals for short-term traffic flow prediction, which fails meet requirements signal control based on cycle signals, this paper proposes an improved long memory-based method periodic volume prediction. The presented in study involves improvements Long Short-Term Memory (iLSTM) and Bidirectional (iBiLSTM) models, leading construction iBiLSTM-iLSTM-NN model. This model incorporates spatial data from surrounding intersections employs fitting techniques establish correlation between queue length volume. Subsequently, a predictive is developed correlation, enabling reliable forecasting future volumes within given cycle. Additionally, actual intersection collected simulation analysis. results indicate that prediction error influenced by different characteristics such as peak, off-peak, normal periods, well inbound lanes. Different parameters have noticeable impact model’s performance, with smaller batch sizes more stable models. By comparing performance models using various evaluation metrics, finds proposed exhibits most performance. research findings can be applied rapidly predict several periods instantaneous at end red phase, providing reliable, accurate, timely urban control.