01662736

Component

Human visual inspection is the most common evaluation technique to estimate roadway conditions. Highway agencies use statistical tools to conjecture overall conditions of highway networks based on a limited number of surveys. This approach requires a high level of coherence and accuracy during the sampling process. Unfortunately, human visual inspections are error-prone activities that may bias the results. This paper presents a two-phase sampling strategy to account for the uncertainty inherent to visual inspections. The proposed design allows to evaluate inspectors´ performance using parameters such as Probability of Detection (POD) or False Alarm Rate (FAR). Receiver Operating Characteristic (ROC) and Precision Recall (PR) graphs are used to illustrate the results. An effective and straightforward method is developed to correct the estimates based on performance results. The methodology is applied to Wisconsin highways data demonstrating its ease of use and efficiency.

This paper was sponsored by TRB committee AHD10 Standing Committee on Maintenance and Operations Management.

18-02531

English

16p

2018

Transportation Research Board 97th Annual Meeting

Digital/other

Figures; References; Tables

Evaluation and assessment; False alarms (Security); Highway maintenance; Improvements; Inspection; Reliability (Statistics)

Highways; Maintenance and Preservation; Planning and Forecasting

Transportation Research Board Annual Meeting 2018 Paper #18-02531

TRIS, TRB, ATRI

Jan 8 2018 10:36AM