A Vibration-Based Methodology to Monitor Road Surface: A Process to Overcome the Speed Effect.

Area of Science:

• Civil Engineering
• Transportation Engineering
• Data Science

Background:

• Effective road pavement monitoring is crucial for timely maintenance and urban infrastructure management.
• Current large-scale monitoring methods in urban areas can be costly and slow, failing to capture pavement degradation dynamics.
• Existing vibration-based road monitoring technologies are often speed-dependent and inadequate at lower vehicle speeds.

Purpose of the Study:

• To statistically analyze the impact of monitoring vehicle speed on road pavement condition assessment.
• To develop and introduce a novel methodology to mitigate the speed effect in road monitoring data.
• To predict road surface damage severity using a machine learning approach based on speed and pavement deterioration index.

Main Methods:

• Statistical analysis of road condition data collected by instrumented taxi vehicles at various speeds.
• Development of a machine learning model to establish decision boundaries for damage severity prediction.
• Utilizing vehicle speed and a pavement deterioration index as primary input parameters for the predictive model.

Main Results:

• The study quantifies the influence of vehicle speed on the accuracy of road pavement condition monitoring.
• A machine learning process was successfully implemented to overcome speed-related measurement biases.
• The developed methodology achieved over 80% accuracy in predicting road damage severity levels in real-time.

Conclusions:

• The proposed machine learning approach offers a user-friendly and efficient solution for real-time road pavement condition assessment.
• This method effectively addresses the challenge of speed dependency in vibration-based road monitoring systems.
• The findings support the integration of smart vehicle data for enhanced road network maintenance strategies.