Genetic-Algorithm and Clustering Technique Based Traffic Weight Data Repairing Models

摘要: Weigh-In-Motion (WIM) data are frequently corrupted or not usable due to problems in sensor quality, calibration, environment, a combination of any these factors. As result, the traffic weight usually fall following failure cases: fluctuated data, two peaks shifted, one peak shifted and high percentage overweight trucks. The belonging cases dominate large portion collected from WIM stations Arkansas, possibly other states as well. Without accurate stations, various engineering policy tasks analysis, pavement design, transportation planning can be compromised. This paper introduces methodologies repair bad data: Weight Data Repairing Model (WDRM) Genetic Algorithm (GA) based WDRM. concept WDRM is add remove number vehicles each bin so that newly modified set pass quality control. In GA WDRM, used core control then applied optimization process make closer original have better performance than generated Validation sets with pre-determined level confidence necessary for user rely on repaired inputs to, such Mechanistic Empirical Pavement Design Guide (MEPDG) software. Therefore, validating model (WDVM) clustering techniques established demonstrate made useful analysis. this validation model, pre-cluster first up time attribute month season information data. A into qualified by using gross good