Neural Network Optimal Model for Classification of Unclassified Vehicles in Weigh-in-Motion Traffic Data

A weigh-in-motion (WIM) system has the capability to perform on-site vehicle classifications based on FHWA schema. However, WIM datasets often contain a significant portion of vehicles that could not be classified into any 13 classes by devices. Possible reasons for classifier failing classify these are tailgating, lane changing, traffic congestion, and equipment malfunction. Analysis unclassified was performed with WIM-recorded data. neural network model established determine appropriate allocations classes. Since number is fairly high, will help improve accuracy truck data thus pavement design. Video records streams an interstate section from nearby station were used identify causes misclassifications. The optimal developed through algorithm design, processing, training, validation, robustness analysis, verification video records. It found effective in allocating able reclassify had non-zero attributes high accuracy. provides useful tool properly specified can applied allocate appropriately design would potentially increase benefit reduce cost reliable realistic designs.