TRAFFIC FLOW PREDICTION BASED ON VANET DATA BY COMBINING ARTIFICIAL NEURAL NETWORK AND GENETIC ALGORITHM

Abstract

In many developing countries, predicting traffic flow is one of the solutions to prevent congestion on highways and routes, and the intelligent transportation system is considered one of the solutions to problems related to transportation and traffic. Knowledge of the predicted situation for traffic flow is essential in traffic management and informing passengers. This research presents a short-term intelligent transportation traffic flow forecasting model, which first examines how traffic forecasting can improve the performance of intelligent transportation system applications. Then the method and basic concepts of traffic flow forecasting are introduced, and the two main categories of forecasting, statistical models and machine learning-based forecasting methods (supervised and unsupervised) are discussed. Finally, a method based on machine learning using a genetic algorithm is Presented. The prediction was used as a powerful method for the mathematical modeling of traffic data in the proposed genetic algorithm method to select important traffic data features and neural networks for classification. The simulation and results presented in this research show a 3 percent improvement in traffic flow prediction with the proposed method, which uses SVM as a classifier in the primary method, and the simulation of this method has output a value of 93.6, But the suggested method has an output of 96.6