Hits:

Indexed by:Journal Papers

Date of Publication:2020-07-01

Journal:STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL

Included Journals:SCIE

Volume:19

Issue:4

Page Number:987-1002

ISSN No.:1475-9217

Key Words:Dam health monitoring; hydrostatic load; temperature effect; kernel extreme learning machines; concrete gravity dams

Abstract:Statistical models have been used for dam health monitoring for many years and have achieved some successful applications. In the statistical model, dam structural response is related to external environmental factors such as reservoir water level, temperature, and irreversible time deformation. For concrete dams, the structural response is affected greatly by the ambient temperature. Therefore, in order to establish a more reliable dam health monitoring model, the temperature effect and modeling method should be further studied. This article presents a dam health monitoring model using measured air temperature for temperature effect simulation based on kernel extreme learning machines. The temperature effect is simulated by long-term air temperature data, and the nonlinear relationship is modeled by kernel extreme learning machines, which is an intelligent machine learning technique with high learning speed and good generalization performance. The proposed dam health monitoring model is verified on a real concrete gravity dam with efficient safety monitoring data. Results show that the proposed approach with a variable set recommended for concrete dam behavior prediction is feasible.