Research On Traffic Signal Timing Optimization Based On Traffic Flow Prediction

Traffic congestion has always been a bottleneck restricting traffic development.Due to the lack of long-term transportation planning in most cities,the current transportation hardware facilities cannot meet the demand for road capacity after the significant increase in the number of vehicles.At the same time,hardware facilities are difficult to change in the short term,and the cost of retrofitting is huge.How to improve the road capacity under the existing conditions and use advanced methodologies has become an urgent problem to be solved.After studying the theoretical methods of short-term traffic flow and traffic signal control,this paper finds that although there are many methods for short-term traffic flow prediction and traffic signal timing,there is a lack of connection and no perfect system..In terms of short-term traffic flow prediction,due to the inherent defects of the current method,it is not able to accurately predict the traffic flow.In terms of traffic signal timing,most of the current signal timing schemes are passive adjustments for traffic flow changes.Lack of active response,its regulatory capacity has a certain lag.Therefore,in this study,the traffic flow prediction and signal timing are first considered as two related modules in a system.From the short-term traffic flow prediction,how to adjust the signal timing scheme is studied.The main research contents mainly include the following parts:(1)Constructing a short-term traffic flow forecasting model.Based on the data obtained by the Minnesota Traffic Management Center,a stacking-based short-term traffic flow combination forecasting model different from the traditional single forecasting model was constructed.The accuracy of the model increased by 13.963% compared with the single model.(2)Improve the proposed model.In the basic forecasting model,only the factors of the transportation system itself are considered,but there are many external factors that affect traffic.Considering the influence of weather on traffic flow in the improved model,after correlating the weather data with the traffic flow,the temperature,wind speed and humidity with the highest correlation with the flow rate and the weather state after one-hot coding are selected as the new model.The input characteristics,the model accuracy rate is further improved.In practical application,a time sliding window is added to the model,and the training set is dynamically updated according to the cycle to realize real-time update prediction of the traffic flow.(3)Signal timing optimization.Based on the predicted traffic flow,the timing of the signal control scheme is optimized.The idea is to use the predicted value of the predictive model to calculate the optimal index value of the traffic intersection in the next period,and search for the next moment before the actual traffic flow.Excellent traffic cycle.The experimental results show that compared with the traditional short-term traffic flow prediction model,the new model has higher accuracy and considers the influence of weather events on the traffic flow.The sliding model constructed thereafter is applied in practical engineering.It also has practical significance;at the same time,the improved timing scheme has better traffic capacity than the original timing scheme.This scheme has certain enlightening significance for practical application in solving the hysteresis of signal timing.