A Mixture Model for Predicting Patterns of Spatial Repeatability in Heavy Vehicle Fleets

This paper presents a statistical heavy vehicle fleet model for predicting patterns of statistical spatial repeatability (SSR), i.e., the mean pattern of dynamic tyre force applied to a section of pavement by a large number of similar vehicles. Data from a Multiple Sensor Weigh-in-Motion (MS-WIM) system, collected for a sufficiently large number of vehicles, can be used to identify and measure SSR. A Bayesian analysis technique is used to infer the statistical distributions of fleet properties, given measured axle forces from MS-WIM data. The topic is introduced with the simple example of using the technique to predict distributions of axle weights, based on simulated MS-WIM measurements. The statistical Mixture model presented herein builds on previously presented models to add the necessary complexity and flexibility to represent the bimodal nature of truck fleets (e.g. the presence of both unladen and laden vehicles in the population). The model is numerically validated using simulated MS-WIM data to condition the Bayesian analysis.