loading page

Enhanced Multi-objective Optimization Model for Bridge Performance Assessment and Prediction, Based on Improved PCA, K-means Clustering, and Kaplan-Meier Survival Algorithm
  • +3
  • Chengzhong Gui,
  • Zhi Duan,
  • Zuwei Huang,
  • Zhiguo Sun,
  • Wei Qiao,
  • Yu Cheng
Chengzhong Gui
Institute of Disaster Prevention

Corresponding Author:[email protected]

Author Profile
Zhi Duan
CCCC Highway Consultants Co Ltd
Author Profile
Zuwei Huang
Ministry of Transport of the People's Republic of China
Author Profile
Zhiguo Sun
Institute of Disaster Prevention
Author Profile
Wei Qiao
CCCC Highway Consultants Co Ltd
Author Profile
Yu Cheng
CCCC Highway Consultants Co Ltd
Author Profile

Abstract

The research proposes a hybrid algorithm model that combines model-driven and data-driven approaches for the direct application of bridge health monitoring technology in bridge management. This comprehensive study encompasses a series of analytical techniques and methodologies to build a multi-objective optimization model for bridge performance assessment and prediction. It focuses on the processing of multi-source heterogeneous data, selection of key sub-parameters using Principal Component Analysis (PCA), enhanced K-means clustering analysis, determination of structural component target thresholds, time-dependent survival probability analysis, regression fitting, and timing prediction of the bridge system for both the components of double-layer truss arch bridge and the bridge system. The initial phase of the study concentrates on the diversification and decentralization of monitored data from various sources, integrating and cleaning data obtained from different sources to ensure data quality and consistency. PCA technique is applied to identify key sub-parameters that have significant impacts on the performance of structural components. Enhanced K-means clustering analysis is carried out to effectively group and classify the identified key sub-parameters. Numerical simulations, including structural nonlinear analysis, are conducted to determine the target thresholds of bridge structure, providing important benchmarks for performance evaluation. Finally, a multi-parameter regression model is used to evaluate and update the performance of the bridge structure, taking into account survival probability (using the Kaplan-Meier method), maintenance history, and material deterioration to estimate the most critical time for the bridge system. A case study is conducted to validate the suggested comprehensive algorithms for a double-layer truss arch combination bridge, which contributes to enhancing performance evaluation and predicting the most critical time for structural components and the bridge system in the bridge management and maintenance practices. It should not be ignored that, the accuracy and reasonability of bridge structure system performance evaluation and prediction depend largely on the selection of target thresholds.
28 Mar 2024Submitted to Engineering Reports
29 Mar 2024Submission Checks Completed
29 Mar 2024Assigned to Editor
29 Mar 2024Review(s) Completed, Editorial Evaluation Pending
05 Apr 2024Reviewer(s) Assigned
21 Jun 2024Editorial Decision: Revise Major
30 Aug 20241st Revision Received
03 Sep 2024Submission Checks Completed
03 Sep 2024Assigned to Editor
03 Sep 2024Review(s) Completed, Editorial Evaluation Pending
07 Sep 2024Reviewer(s) Assigned
21 Sep 2024Editorial Decision: Accept