Research On Technologies Of Information Fusion In The Field Of Urban Transportation

With the continuous development of China's urbanization process,urban traffic problems have become increasingly prominent.In the future,intelligent transportation system will be the most important method to solve the urban traffic demand.The realization of intelligent transportation system requires real-time perception and monitoring of road traffic conditions.Fortunately,with the continuous development of mobile communication,satellite positioning,Internet of things,big data and other technologies,GNSS,RFID,microwave,geomagnetism,video and other acquisition methods have been widely used in the field of urban traffic information perception.The diversity of collection equipment,on the one hand,expands the channels of information acquisition and the breadth of information collection;on the other hand,it also causes problems such as large amount of information,strong heterogeneity,data conflict,etc.,which brings difficulties for the effective use of traffic information.The continuous development of information fusion technology provides new ideas and methods to solve this new problem.At present,most of the existing research results on information fusion model inherit the application environment and technical characteristics of the military field,but they are not completely consistent with the urban traffic field in terms of function definition,information characteristics and application objectives,which are difficult to apply to all kinds of applications in the urban traffic field.Therefore,based on the research background of urban traffic field,this paper carries out the research on Key Technologies of information fusion,puts forward the fusion model and method suitable for the information characteristics and application needs of this field,in order to solve the scientific problem of the effective fusion application of multi-source information in urban traffic field.The innovative work of this paper mainly includes the following aspects:1.Research on multi-source information fusion technology with the same attribute.Aiming at the problem of data conflict between multi-source information which affects the accuracy of fusion,a KDS-R model suitable for the same attribute information fusion is proposed.The model combines the structure of Kalman filter with the algorithm of D-S evidence theory,and introduces the static and dynamic reliability analysis of evidence to improve the classical algorithm of D-S evidence theory.The experimental results of simulation data and real data show that the model can effectively deal with data conflicts.Compared with the classical D-S evidence theory algorithm,the fusion result is closer to the actual situation.2.Research on multi-source information fusion technology of complex attributes.Aiming at the problem that it is difficult to map and fuse the information with complex attributes,this paper focuses on two traffic parameters,speed and flow.Based on the measured traffic data,the relationship between them is studied and analyzed.A CANN fusion model based on association rules and BP neural network is proposed.By constructing and optimizing the mapping relationship between speed and traffic,the fusion calculation between them is realized.The experimental results show that the variance of the prediction curve based on this model is smaller than that based on single flow information,and it is closer to the measured flow curve.3.Research on traffic prediction model based on multi-source information fusion.Based on the comprehensive analysis of time and space characteristics of traffic information,a short-term traffic flow prediction model based on WNN and a bus arrival time prediction model based on information fusion are proposed.The short-term traffic flow prediction model focuses on the spatial attributes of traffic information.Through the correlation analysis of speed and flow,the virtual cross-section is fitted,and the granularity of road flow description is further refined.KDS-R model and CANN model are used to complete the prediction combined with wavelet neural network technology.The bus arrival time prediction model focuses on the time attribute of traffic information,combines the traffic flow information to judge the road traffic conditions,and then uses KDS-R model to complete the arrival time prediction.The experimental results show that the prediction model of short-term traffic flow based on WNN converges faster than the prediction model based on BP neural network,the prediction results are closer to the actual curve,and the error is reduced by about 50%;the prediction model of bus arrival time based on information fusion effectively improves the problem of low sensitivity of single data to road congestion which affects the prediction accuracy.The prediction effect is better than the prediction results based on the front car data and the historical average data.4.Design and implementation of bus big data platform based on information fusion.Based on the above research results,the bus big data platform based on information fusion is designed and implemented,which solves the key technical problems such as data conflict and information redundancy in the complex urban traffic environment,and provides support for the bus information release service in the big data platform.Based on the actual operation data of bus No.82 in Pudong New Area of Shanghai,the operation effect of the bus arrival time module in the big data platform is shown in detail.The prediction accuracy reaches 96%,which is superior to the management standard requirements of Shanghai public transport industry,and provides a guarantee for the comprehensive construction of public transport information service in Pudong area.