TRAVEL TIME MEASUREMENT BY VEHICLE SEQUENCE MATCHING METHOD USING GENETIC ALGORITHM

摘要

We studied a method of identifying vehicles using vehicle height data obtained from ultrasonic vehicle detectors. We previously proposed vehicle matching based on vehicle sequences (two or more vehicles) instead of one-to-one matching, because many similar vehicles exist on the road. We developed a Genetic Algorithm method based on traffic characteristic, and we describe an improved method of evaluating vehicle sequences using Levenshtein Distance, also called the edit distance, corresponding to expression of vehicle behavior between upstream and downstream sides. We report on the accuracy of vehicle identification and found that travel time measurement was improved.