A Data-Mining Interpretation Method of Pavement Dynamic Response Signal by Combining DBSCAN and Findpeaks Function.

Area of Science:

• Civil Engineering
• Data Science
• Materials Science

Background:

• Asphalt pavement strain data under heavy traffic often shows fluctuations and anomalies due to complex environments.
• Effective real-time diagnosis of anomalous data is crucial for accurate pavement monitoring and maintenance.

Purpose of the Study:

• To develop a real-time anomalous-data diagnosis system for asphalt pavement.
• To effectively extract dynamic strain features from sensor data.
• To improve the efficiency of data processing and information utilization in pavement monitoring.

Main Methods:

• Data denoising using low-pass filters.
• An improved DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm with the K-Dist method for anomaly detection.
• A refined findpeaks function for adaptive feature extraction of denoised data.

Main Results:

• The enhanced DBSCAN algorithm effectively detects abnormal data in road dynamic response signals.
• The refined findpeaks function accurately identifies peak values and strain signal peaks, even for complex multi-axle lorries.
• The proposed method demonstrates effectiveness in handling large datasets and extracting critical features.

Conclusions:

• The developed method provides an effective approach for real-time anomalous data diagnosis in asphalt pavement monitoring.
• Accurate feature extraction using DBSCAN and findpeaks enhances pavement management and maintenance strategies.
• This research contributes to more efficient data processing and utilization in the field of pavement engineering.