Big Data Analytics and Structural Health Monitoring: A Statistical Pattern Recognition-Based Approach

Area of Science:

• Engineering
• Data Science
• Civil Engineering

Background:

• Advancements in sensor technology have led to big data in structural health monitoring (SHM).
• Data-driven methods offer opportunities for long-term SHM strategies using vibration data.
• Challenges in SHM include complex feature extraction and decision-making with big data.

Purpose of the Study:

• To propose an efficient strategy for feature extraction and classification in SHM.
• To address limitations of current methods in handling big data and high-dimensional features.
• To assess the effectiveness of a novel approach for damage detection in bridges.

Main Methods:

• Feature extraction using autoregressive moving average (ARMA) modeling.
• Feature classification employing an innovative hybrid divergence-based method.
• Validation using vibration data from a cable-stayed bridge.

Main Results:

• The proposed ARMA modeling and hybrid divergence-based method effectively extracts features.
• The hybrid method demonstrates high efficiency in classifying features for damage detection.
• Successful damage detection was achieved in a big data context for the cable-stayed bridge.

Conclusions:

• The combined ARMA modeling and hybrid divergence-based method is effective for SHM.
• This strategy successfully overcomes challenges associated with big data in structural monitoring.
• The approach offers a viable solution for long-term, data-rich structural health assessment.