Procedure for Statistical Categorization of Overweight Vehicles in a WIM Database

Recently observed increases in the numbers of permitted and illegal overweight trucks travelling over U.S. highways have raised concerns over their contributions to the reduction in the service lives of pavements and bridges and the costs of maintaining, upgrading and replacing the highway infrastructure system. Cost allocation studies are used by transportation officials to help their asset management processes and to establish truck traffic and permitting policies taking into consideration information on the composition of overweight trucks and their permit categorization. In recent years, cost allocation studies have heavily relied on data assembled by weigh-in-motion (WIM) systems which provide information on traffic counts, truck axle configurations and weights for various highway classes and economic regions. However, WIM data by themselves do not provide information on the numbers of illegal overweight trucks because many of the overweight trucks may have been issued permits that allow them to operate on a yearly basis or on a trip-by-trip basis. The object of this paper is to develop a data mining procedure to identify and classify overweight vehicles whose axle weights were collected by WIM systems into different permit and illegal categories. The first step of the proposed data mining procedure establishes a set of rules that are satisfied by different types of permit trucks. These rules are inferred from a review of available permit databases. In a second step, a search algorithm is used to check each vehicle in a WIM database and identify whether it violates any of a jurisdiction’s legal weight limits. Subsequently, every overweight truck’s axle weights, axle spacings, total length and gross vehicle weight are checked to verify whether these characteristics match any of the criteria established when setting up the data mining rules. The overweight vehicles are separated based on the likelihood of having been issued a particular type of permit or if they are potentially illegal. The validity of the algorithm is demonstrated by analyzing truck data collected at a WIM site in upstate New York. The algorithm’s output showed reasonable agreement when compared to the results of a truck survey performed by the New York State Department of Transportation (NYSDOT). A parametric analysis is executed to assess the sensitivity of the results to the level of accuracy of the WIM system. It is concluded that the algorithm produces statistically robust overweight truck and permit categorizations which can eventually help highway agencies establish rational permit issuance policies, weight enforcement strategies, and cost allocation studies.