Research On Key Technologies Of Route Optimization For Urban Road Network Based On GPS Data

The widespread utilization of global positioning system(GPS)devices has led to the generation of large amounts of urban spatiotemporal data,which are crucial for executing traffic planning tasks(e.g.,trip time estimations and optimal path searches).Furthermore,individuals can utilize GPS devices to collect floating vehicle positioning data,which are time-interval data that describe moving vehicles using parameters such as the current location,time,speed,and direction.These devices exhibit a low input cost,and they are widely utilized in vehicles such as taxis,buses,and trucks.Over the past 10 years,the utilization of probe vehicle positioning data-collection technology has been widely promoted in the major cities of China;thus,the accumulation of a large amount of positioning data that comprehensively describes urban road networks has occurred.With regard to accurately estimating urban road travel times,the utilization of probe vehicle data offers crucial research opportunities and practical opportunities;thus,the vehicle routing that characterizes urban road networks is optimized.By utilizing GPS-based applications,road users can identify optimal alternative paths,and by reducing the number of cars that utilize the network at any given time,which can ease congestion,these applications can enhance the quality of users’trips.This research was conducted in four stages.(1)The development of a map-matching algorithm that is based on the hidden Markov model and the parallel Viterbi algorithmWe studied the map-matching algorithm that is based on floating vehicle positioning data.The construction of the map-matching algorithm is based on the hidden Markov model,and it considers the driving direction,which is a distance-based parameter;furthermore,to solve the optimal state sequence,the parallel Viterbi algorithm is utilized.The aforementioned method is verified by the Changchun road network and GPS data.The results indicate that the method exhibits high matching accuracy,even when the sampling interval is large,and that the average computational time is quite low.This chapter provides a data basis for subsequent travel-time estimation.(2)Estimation of urban road section travel times using GPS dataThis study utilized a large sampling interval,through which the researchers evaluated the travel time of a floating vehicle without a GPS anchor point.Therefore,the GPS anchor points pertaining to the upstream and downstream were identified,and to match the characteristics of the target floating car to the most similar historical floating car,the k-nearest neighbor rule was utilized.This historical floating car was then utilized as the target anchor point.Subsequently,the virtual registration point pertaining to the target floating vehicle that characterizes the target section was constructed,and the missing registration point data were repaired.Moreover,using the virtual registration points and actual registration points as inputs,an enhanced PSO-XGBoost regression model was utilized;to estimate the road travel time.Finally,using a main road that exists in Changchun,which was utilized as an example,the method was tested and evaluated,and based on the preceding method,the network road travel time estimation was constructed.The results indicate that the method proposed herein exhibits high estimation accuracy and robustness.(3)Research on the urban point-point shortest path search algorithm that considers turning type and signal coordination controlWith regard to the road resistance input of the network structure,the road travel time is utilized.By considering the waiting time mutation that is occasioned by phase waiting,which occurs at the signalized intersections pertaining to different turning types,the waiting time model for signalized intersections is constructed.Moreover,to shorten the vehicle travel time that characterizes an urban road network,a trunk-coordinated green wave band can be utilized;therefore,to optimize the intersection wait time model,we introduced the phase difference parameter.Subsequently,to characterize the steering relationship between sections and the phase sequencing of the corresponding intersections,an enhanced star table structure was utilized.Thus,the signalized intersection wait time model was superimposed on the iterative,deepening A*algorithm,which led to the construction of the AICIDA*algorithm(i.e.,an algorithm that considers the coordinated control of trunk lines).The application of these algorithms to the road network in Changchun indicated that compared with the traditional algorithm,the AICIDA*algorithm was effective at calculating signalized intersection wait times,and that the AICIDA*algorithm was superimposed on the waiting time model.With regard to the AICIDA*algorithm,the total cost pertaining to the path search was low,and the search speed was fast.(4)Research on the multi-rule sequenced path optimization algorithm considering city point of interestBased on point-point shortest path search algorithm research,the path optimization algorithm considering city point of interest(POI)multi-rule sequencing access restrictions is studied.To represent all POI sequences that meet the travel rules,the vertex activity graph,which indicates the activity on vertex network(AOV)network and the generalized rule tree(T_R)are utilized.Thus,a novel route optimization problem is proposed,namely the turning cost constrained multi-rule partial sequenced route(TCC-MRPSR)problem,which considers turning cost,and two POI multi-rule sequencing path optimization algorithms that consider turning cost are designed.Finally,using the road network of Changchun City for simulation verification,the analysis indicates that compared with traditional algorithms,the proposed algorithm can provide a shorter path under the constraint of travel rules,and that the operation time is short enough for application.