Pergamon zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA TranspnRes.-B, Vol. 31, No. 2, pp. 141-157, 1997 0 1997 Elsevier Science Ltd All rights reserved. Printed in Great Britain 0191-2615/97 $17.00+0.00 PII: SO191-261S@5)OOO12-4 PARAMETER OPTIMIZATION METHODS FOR ESTIMATING DYNAMIC ORIGIN-DESTINATION TRIP-TABLES HANIF D. SHERALI, NAMITA ARORA Department of Industrial and Systems Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061-0118, U.S.A. and ANTOINE G. HOBEIKA Department of Civil Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061-0105, U.S.A. (Received 7 September 1994; in revised form 6 February 1996) Abstract-Dynamic origin-destination tables help in on-line control of traffic facilities and, con- sequently, are of significant use in alleviating traffic congestion. Such tables find useful applications in the contexts of Advanced Traffic Management Systems and Advanced Traveler Information Systems. This paper considers theestimation of split parameters that prescribe an origindestination trip-table based on dynamic information regarding entering and exiting traffic volumes through an intersection or a small freeway segment. Two models are developed and motivated for this problem, one based on a least-squares estimation approach and the other based on a least absolute norm approach. Both models enhance existing dynamic origindestination trip-table estimation models in that they also consider freeway segments having differing time-dependent transfer lags between different pairs of entrances and exits. A projected conjugate gradient scheme is employed for solving the constrained least-squares problem and is compared against a standard commercial software. The least absolute norm estimation problem is posed as a linear programming problem and is also solved using a commercial software for the sake of comparison. Computational results are presented on a set of test problems using synthetic as well as realistic simulated data, involving the determination of origindestination trip tables for both intersection and freeway scenarios, in order to demonstrate the viability of the proposed methods. These results exhibit that, unlike as reported in the literature based on previous efforts, properly designed parameter optimization methods can indeed provide accurate estimates in a real-time implementation framework. Hence, these methods provide competitive aitematives to the iterative statistical techniques that have been heretofore used because of their real-time processing capabilities, despite their inherent inaccuracies. 0 1997 Elsevier Science Ltd. All rights reserved I. INTRODUCTION Dynamic origindestination (O-D) trip-tables provide key information for effective on-line route guidance in the contexts of Advanced Traffic Management Systems (ATMS) and Advanced Traveler Information Systems (ATIS). Whenever events occur that cause congestion such as accidents, road maintenance work, or chemical spills, fast and efficient traffic control and diversion strategies are required. A crucial information in developing successful control and diversion strategies in such situations is the knowledge of the destinations of travelers. O-D trip-tables are one way of representing flows through a traffic facility. An GD trip-table is a two-dimensional matrix of elements whose cell values represent the number of trips made between various O-D zone pairs in a given region. For implementing reliable on-line traffic control strategies, it is advisable that the O-D trip-table for the traffic facility under consideration be updated frequently, in keep- ing with changes in the traffic pattern. Dynamic O-D estimation procedures address the problem of estimating such trip-tables in real time. The goal here is to estimate the proportion of traffic that flows between each I41