Increasing Reliability of Participatory Sensing for Utility Pole Condition Assessment Using Fuzzy Inference

Abstract

Aging infrastructure has become a safety issue for local communities. For example, aging and deteriorating utility poles in strong winds have a high risk of falling onto roads, adjacent houses, or vehicles, which could cause traffic jams, power outages, property damage, or casualties. To prevent such accidents, the infrastructure condition needs to be monitored on a regular basis, thereby facilitating proactive maintenance and repair. However, available monitoring resources are limited for inspecting numerous existing infrastructures in a timely manner, which hinders obtaining up-to-date records representing the current condition status. As an alternative monitoring method, participatory sensing has the potential for infrastructure inspection, leveraging the prevalence of citizens’ smartphones as a ubiquitous sensing device. Nonetheless, the reliability of crowdsourced data for infrastructure monitoring and how to improve it have not been investigated fully, although participatory sensing increasingly has been adopted in many studies. Because citizens generally do not have expertise in infrastructure condition assessment, a lack of understanding of the crowdsourced data reliability prevents participatory sensing from being implemented in practice. To advance the understanding of the crowdsourced data reliability for infrastructure assessment, this study investigated the discrepancy in infrastructure assessment results by citizens and by an expert. Wood utility poles were selected as a target infrastructure. This study investigated a way to reduce the deviation between the expert’s and the citizens’ responses through a fuzzy inference system along with particle swarm optimization and pattern search algorithms. The experimental results showed that the proposed fuzzy inference system reduced the evaluation error by 21.18%. The findings of this study have the potential to fill the knowledge gap for enhancing participatory sensing in infrastructure monitoring, thereby promoting future study to enhance its applicability for citizen-driven urban resilience.