Instantaneous follower-vehicle speed prediction under car-following conditions for intelligent transportation system applications: a rime optimization algorithm–based deep hybrid model

This study presents a deep learning framework for predicting the instantaneous speed of follower vehicles under car-following conditions, with potential applications in Intelligent Transportation Systems (ITS). The proposed model combines Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and an attention mechanism to capture spatiotemporal dependencies and highlight critical historical time steps. To improve model performance, the Rime Optimization Algorithm (RIME) automatically tunes key hyperparameters in the CNN-BiLSTM-Attention model. Valid car-following samples were extracted from the publicly available Next Generation Simulation (NGSIM) dataset using a rule-based follower–leader matching procedure. The model uses follower speed, leader speed, and space headway as inputs to predict future follower speed. Data preprocessing includes the Augmented Dickey-Fuller (ADF) test, first-order differencing, and normalization to improve data stationarity. The influence of historical sequence length is examined under one-step-, five-step-, and ten-step-ahead prediction tasks. Experimental results show that the proposed model achieves strong overall performance compared with Autoregressive Integrated Moving Average (ARIMA), Support Vector Regression (SVR), and ablation variants. Compared with Particle Swarm Optimization (PSO) and Sparrow Search Algorithm (SSA), RIME provides better overall error performance and a favorable balance between prediction accuracy and computational efficiency. Event-level validation across multiple car-following groups demonstrates reliable prediction during rapid acceleration and deceleration under dynamic car-following interactions. These findings support the proposed model for short-term follower-speed prediction and potential applications in traffic safety and vehicle decision-support within ITS.