Research On Model And Algorithm Of Traffic Flow Control On Intersection Under The Background Of Intelligent Transportation

To strive to develop intelligent transportation is the main direction for constructing an innovative development system of transportation in nowdays.Intelligent transportation is a qualitative leap for the new era of intelligent transportation system which makes internet of things,cloud computing,big data,wireless sensor,'internet +',artificial intelligence and some other advanced technologies into use aiming at gradually improving the scientific&intelligence level of the whole transportation system.So that better services can be provided to road users and it will be much more convenient to live.In the future of traffic development,the construction of transportation infrastructure should be appropriately developed on the basis of correctly assessing the level of regional economic and social development due to the limitation of natural resources such as land.At the same time,we should make full use of new ideas and technologies,and vigorously promote scientific management of transportation to form the basis of the continuous development of the control and guidance of traffic.With the construction of intelligent transportation,this paper makes further research on traffic flow control technologies of the road intersection,which is the key point of traffic control and guidance.And the research involves the following specific contents:(1)The existing traffic flow prediction models cannot extract the temporal-spatial features of traffic flow data and the performance of them is susceptible to external interference factors.To solve this problem,an innovative short-term traffic flow prediction model based on deep learning is proposed in this paper.Characteristics of Convolutional Neural Network(CNN)and Support Vector Regression(SVR)are combined together in this model,which uses CNN to extract features of traffic flow data in bottom network layer and put them into SVR to for flow predictions.To verify the effectiveness of the proposed model,experiments are carried out using actual traffic flow data and the results demonstrate that the proposed model is an effective traffic flow prediction model which has higher accuracy than traditional ones.(2)On the basis of analyzing theories and applications of typical timing methods,three optimizing indexes for dynamic signal timing of intersections are confirmed and a multi-objective programming model is proposed.In order to solve the equation better and obtain a best timing plan,bionic dragonfly method is made into use and modified by Gaussian hybrid mutation operators to insure the diversity of solution set.Dynamic external file is built to make sure that the optimal solution set is dynamically balanced.Benchmark functions are utilized in trial stages to verify the effectiveness of the modified algorithm.Simulations about the proposed dynamic timing model and solving algorithm are conducted on vissim platform.Simulation results generated by the proposed method are compared with Webster and the comparison results show the superiority of the proposed method.(3)A cooperative map-matching(CMM)algorithm is proposed based on the analyzing of positioning principle of global positioning system(GPS),vehicular dead reckoning(DR)and extended Kalman algorithm.In the investigation,information obtained by GPS and DR is integrated first by using extended Kalman algorithm and the integrated result is treated as the initial position of CMM algorithm.Then,sharing and exchanging of vehicle information is fulfilled by using short range communication.The precision of localization is improved by taking advantage of road constraints based on electronic map.In order to verify the effectiveness of the proposed algorithm,Experimental results demonstrate that the proposed cooperative CMM algorithm has higher positioning accuracy compared with traditional vehicle positioning algorithms.(4)A vehicle risk assessment model is proposed based on the analyzing of basic knowledge and design of control flow of vehicle collision avoidance,in addition,the assessment of intersection conflict situation and the calculation of relevant parameters.A optimal control model of vehicle collision avoidance at non-signal intersection is proposed based on analyzing the mathematical expression of optimal control problem and building vehicle motion state equation.Genetic algorithm is used to solve the equations,the steps include introducing the flow of genetic algorithm,encoding collision avoidance path and constructing fitness function and the provisions of genetic opration.Finally,experiments simulating different vehicle encounter cases are conducted using models and solving algorithms proposed in this paper.Experimental results demonstrate that the models and solving algorithms are effective.