Unsupervised performance evaluation strategy for bridge superstructure based on fuzzy clustering and field data

Area of Science:

• Civil Engineering
• Structural Health Monitoring
• Data Science

Background:

• Bridge performance evaluation is crucial for effective maintenance.
• Existing methods may lack unsupervised approaches for superstructure assessment.
• Field-measured data offers valuable insights into bridge condition.

Purpose of the Study:

• To propose an unsupervised method for bridge superstructure state assessment.
• To utilize fuzzy clustering and field data for reliable evaluation.
• To establish a foundation for optimal bridge maintenance strategies.

Main Methods:

• Construction of a bridge evaluation index system.
• Application of fuzzy clustering on selected bridge health monitoring data.
• Generation of fuzzy similarity and equivalent matrices.
• Dynamic clustering map creation and statistical analysis for classification.
• Evaluation of unknown bridge states using fuzzy nearness to a sample base.

Main Results:

• Successful implementation of an unsupervised bridge superstructure state assessment method.
• Demonstration of fuzzy clustering's efficacy in classifying bridge health states.
• Validation of the proposed method using data from Nanping Bridge.

Conclusions:

• The proposed fuzzy clustering method provides an effective approach for unsupervised bridge superstructure state assessment.
• This technique aids in determining optimal maintenance strategies by accurately evaluating bridge condition.
• The method shows practical applicability in real-world engineering projects.