Research On Short-term Traffic Flow Forecast And Signal Timing Technology Based On Machine Learning

With the rapid development of China’s economic construction,China’s transportation industry is also undergoing tremendous changes.The concept of smart transportation has gradually penetrated into various industries and fields.The "Outline for the Construction of a Powerful Country" also pointed out that we must vigorously develop smart transportation.The intelligent transportation system is complex in structure and contains various contents.Short-term traffic flow prediction and traffic signal control are its core research contents.However,the accuracy of short-term traffic flow prediction is not high enough,and the prediction results cannot be effectively used in traffic control.Problems such as poor road capacity during peak traffic flow still exist.In order to solve the above-mentioned problems,improve the prediction accuracy of short-term traffic flow on high-speed and urban arterial roads and enhance the road capacity during peak hours,thus strengthening the construction of smart transportation.In this paper,the MPSO optimized relevance vector machine model is used to predict highway traffic flow,compared with the traditional particle swarm algorithm,and multiple prediction models are used to compare the same set of data The prediction and analysis of the data results prove that the improved prediction algorithm can reduce the prediction error to 6.03%,improve the fit of the prediction data,and obtain a better prediction effect.Secondly,the MPSO algorithm optimized LSTM model was established to study the short-term traffic flow prediction of the city’s main roads and the dynamic timing of traffic lights.The data of the main roads in Shenzhen was used to analyze the MPSO-LSTM The effectiveness of the network prediction model was experimentally verified and compared with other classic models.The results show that the improved prediction model has higher prediction accuracy.Finally,a fuzzy controller is designed,which combines the short-term traffic flow data predicted by Shenzhen Fuhua Road and Mintian intersection based on the MPSO-LSTM network model with the controller,designing a set of signal timing algorithm and matching with Webster classic Time method for comparison.It proves that the timing algorithm designed in this paper has higher capacity.