Short-term traffic flow prediction based on improved Chimp algorithm optimized BiGRU

To improve the accuracy of short-term traffic flow prediction, this paper proposes the Improved Chimp Optimization Algorithm (ICOA) to optimize the prediction model (ICOA-BiGRU) for the bidirectional gated recurrent unit (BiGRU). Compared to the initial Chimp Optimization Algorithm (ChOA), the improved algorithm first initializes the population using Hénon chaotic map to increase the population diversity. Second, inertia weights are added to the position update process to balance the global optimization and local exploitation of the algorithm and improve the convergence of the algorithm. Finally, the improved ChOA algorithm is used to optimize the hyperparameters of BiGRU, and the ICOA-BiGRU prediction model is obtained. Four prediction models, long and short-term memory neural network (LSTM), BiGRU, ChOA-BiGRU model, and ICOA-BiGRU model, were trained and tested using short-term traffic flow data, and the prediction results were compared and analyzed using the three metrics of mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE). The results show that the GRU model performs better in short-term traffic flow prediction than the LSTM model; the ICOA-BiGRU prediction model constructed by introducing the improved chimp algorithm has higher prediction accuracy and efficiency compared with other models, and it is able to provide a decision-making basis for the intelligent transportation system.