Traffic data integration in advanced traveler information systems using fuzzy operator logic

The continuously deteriorating performance of transportation systems has created a need for new solutions. Advanced Traveler Information Systems (ATIS), a component of Intelligent Vehicle Highway Systems, is thought to assist travelers in reducing travel times and increasing comfort of traveling. This can be accomplished by using updated, complete traffic information. General data sparsity and variety create a need for efficient data processing from all available sources.;The data integration process is viewed as a tree consisting of two data operations: conversion and fusion. The formalism of fuzzy operator logic and the Dempster-Shafer rule of combination are used as a framework for data integration. The knowledge base has been structured to expose and utilize data redundancy in order to enhance the traffic information quality. The original Dempster-Shafer formalism has been modified to improve fusion of contradictory information. The congestion information desired is retrieved from the knowledge base through a sequence of arithmetic and minimax operations which makes the process of data integration time efficient.;A prototype algorithm for congestion detection has been calibrated from data produced by the INTRAS simulation program. The measure of algorithm performance and the calibration process have been designed for application to real world conditions. The calibration and evaluation results indicate that the algorithm is capable of producing confident answers about congestion conditions for a large number of links when sparse data prevail. The analysis also shows that intelligent data processing may remarkably decrease the need for on-line data by 60-80%.;An implementation of the developed congestion detection algorithm is considered to support incident detection in the networks operating under ATIS. Also the use fuzzy operator logic and the Dempster-Shafer rule of combination for on-line estimation of travel times is discussed.;The primary goal of this research is to investigate the feasibility of using fuzzy logic as a framework for traffic data processing based on an example algorithm of congestion detection in signalized networks. An artificial intelligence approach to data integration, is designed to achieve the maximum benefit by using several data sources and expert rules to interpret and integrate the available data. The additional computational effort required is addressed in the algorithm design.