Road Network Optimization Based On Knowledge Inference

Road network optimization is a practical example of the application of complex network analysis,which is a common means of solving the problems of traffic congestion.At present,traffic congestion not only caused huge waste of time and energy,but also brought huge damage to the environment.In order to meet the traffic demand,some new roads will be added to the road network under the condition that the original road network structure is unchanged,which is a typical problem of road network optimization.This way of road network optimization is in line with the policy of open communities which is promoted by the state in 2016.At present,the research on the open area of closed areas has no quantification,small scale and high cost.This paper uses a variety of data sources to design map matching algorithms to perceive traffic conditions and construct traffic simulation scenarios for simulation.Firstly,the network structure is designed according to different heuristic strategies,and traffic simulation is carried out under different strategies.Although the heuristic method has improved the road network traffic,the improvement is not large,and it is too simple to use any algorithm.On the basis of heuristic method,the regional attribute and open result are represented as rules,and the tree fuzzy inference system is constructed.Through genetic algorithm optimization,the training process is accelerated by bayesian optimization,and the optimal knowledge base is learned.For the first time,the method of knowledge inference is used in the road network optimization.The specific research contents are as follows:(1)Design and implement map matching algorithm.In order to obtain the traffic conditions of an urban road network from GPS track data such as taxis,private cars and buses.First of all,you need to use the map matching technology to match the GPS track points to the corresponding road,and then you can count the traffic speed and traffic flow and so on.Due to the large amount of GPS trajectory data,so the demand of the efficiency of map matching algorithm is very high.The existing methods can't solve the problems well.Therefore,this paper designs a fast map matching algorithm based on grid index.(2)Build a simulation scene.In order to explore the traffic of road network.This paper uses a simulation approach,so we need to build simulation scenes.The main tasks include: Construct road network,estimate OD matrix and generate traffic flow and so on.By SUMO simulator,you can get different road networks.In the different traffic flow,it will produce the corresponding simulation results.(3)Design open strategies and build knowledge base.In order to explore how to open up the community,which can get better results.In this paper,we design a number of different heuristic open methods based on the attributes of the community itself and the traffic conditions around the community.In addition,in order to summarize the generality of open community,this paper presents different community on and off as knowledge base,and determines the community on or off according to the state of input community.The knowledge base is represented as a tree fuzzy inference system(FIS)that is optimized by genetic algorithms.