A New Data-Driven Approach to Traffic Flow Prediction Based on TCOA-BiGRU

Traffic flow is dynamic., uncertain., and nonlinear., which makes it difficult to predict traffic flow accurately. In this paper., a TCOA-BiGRU (Tent Coati Optimization Algorithm-Bidirectional Gated Recurrent Unit) combined optimization prediction model is proposed to forecast traffic movement. Firstly., on the basis of COA (Coati Optimization Algorithm)., TCOA (Tent Coati Optimization Algorithm) is proposed by using Tent chaotic mapping to improve the algorithm. Secondly., TCOA is employed to optimize the parameters of BiGRU (Bidirectional Gated Recurrent Unit) in order to fully leverage its capabilities. The experiment utilizes the public dataset of UK highways as the data source and employs the TCOA-BiGRU combined optimization prediction model for comparison with conventional traffic flow forecasting models. The experimental results show that compared with BiGRU., the RMSE (Root Mean Squared Error) of the TCOA-BiGRU combined optimization prediction model decreases by 8.88%., the MAE (Mean Absolute Error) decreases by 9.36%., and the R2 (Decision Coefficient) reaches 0.93583. Across the three datasets., TCOA-BiGRU demonstrates smaller RMSE and MAE values compared to the other four prediction models., accompanied by a larger R2. These findings serve as empirical evidence supporting the efficacy of the proposed model delineated in this paper., thereby shedding light on the promising prospects of combinatorial optimization prediction models within the domain of traffic flow forecasting.