Research on a Short-Term Traffic Flow Prediction Model Using Enhanced Genetic Algorithms and Neural Network Collaboration

Aiming at the problems of slow convergence speed and easy to fall into local optima in traditional methods for short-term traffic flow prediction, this paper proposes a prediction model based on improved genetic algorithm (IGA) and neural network collaborative optimization. By designing adaptive crossover and mutation probability adjustment mechanisms, hybrid selection strategies, and elite retention strategies, the neural network weights and feature selection process are optimized. Experimental results show that the proposed model achieves a mean squared error (MSE) of 15.3 on a 5-minute prediction scale, which is 22.3 % lower than the standard genetic algorithm (SGA), and the convergence speed is improved by 35 %. The model demonstrates good robustness in complex traffic scenarios, providing an effective tool for real-time decision-making in intelligent transportation systems.