Research On Short-term Traffic Flow Forecasting Based On Optimized And Integrated RBF Neural Network

With the economic development of the society,the number of motor vehicles is constantly increasing,leading to problems such as environmental pollution,traffic accidents and traffic congestion,which have become important bottlenecks restricting urban development.In order to effectively solve urban traffic problems,intelligent transportation systems have emerged at the historic moment.As an important part of the intelligent transportation system,short-term traffic flow forecasting solves the problem of how to provide prediction results accurately and in real time.However,the complexity,non-linearity and uncertainty of short-term traffic flow systems make it difficult for existing prediction models to meet the time and accuracy requirements of intelligent transportation systems.Based on the summary of previous research,this paper detects the original traffic flow data and repairs outliers to ensure the rati onalit y of data analysis and mining.At the same time,it uses radial basis(RBF)neural networks and integrated learning algorithms to establish short-term traffic flow prediction model,which provides theoretical support for traffic guidance and control.The main innovative research work of the paper is as follows:1.A method based on improved FCM algorithm for repairing missing data of traffic flow is proposed.Aiming at the problem of abnormal traffic flow data during the collection and transmission of traffic detectors,an improved method of FCM for repairing missing traffic flow data is proposed.The fuzzy decision theory and simulated annealing algorithm are used to optimize the number of clusters and fuzziness index of FCM.Compare the similarities in traffic flow data,realize the repair of missing data,and provide high-quality data support for subsequent short-term traffic flow prediction research;2.A traffic flow prediction model based on IACO-RBF neural network was established.In view of the shortcomings of traditional neural networks,such as low prediction accuracy and sensitive parameter settings,an ant colony algorithm was introduced to optimize the relevant parameters of the RBF neural network.In order to improve the convergence speed and optimization accuracy,the pheromone update formula and path transition probability in the ant colony algorithm are improved,and the IACO-RBF neural network prediction model is established,and the repaired traffic flow data is used to improve the pred iction accuracy;3.A traffic flow prediction model based on AdaBoost method integrated with IACO-RBF neural network was established.For the prediction of a single RBF neural network,there are problems such as model instability and low generalization abi lity.Using the IACO-RBF neural network prediction model as the base learner of the AdaBoost integrated algorithm,and multiple prediction results jointly determine the final prediction output,thereby improving the stability and prediction accuracy of the prediction model.