Reliability analysis method of gravity dam based on probability-fuzzy-interval hybrid model and improved branch and bound method

Using the traditional probability model to evaluate the service safety of gravity dam needs to clarify the probability distribution of parameters, and it is difficult to accurately measure the service reliability of gravity dams based on the conclusions of non-probability interval model. Therefore, a reliability analysis method of gravity dam based on probability fuzzy-interval-hybrid model and improved branch and bound method is proposed. Based on the prototype, laboratory test results and safety monitoring data, combined with the parameter time-varying model and interval inversion analysis method, a hybrid model for reliability analysis of gravity dam is established, which comprehensively considers random variables, fuzzy variables and interval variables. Using the information entropy method and Karush Kuhn Tucker (KKT) optimization condition decoupling the hybrid model, the reliability index is calculated by the checking point method based on equivalent normalization method (JC method for short). The traditional branch and bound method is improved to search the main failure modes, and the Ditlevsen narrow bound method is used to calculate the system reliability, so as to comprehensively evaluate the overall service safety of the gravity dam. Examples shows that the method can solve the reliability analysis of gravity dams with many uncertain factors, and has a wide range of application. The calculation result is still a probability reliability index. On the premise of conforming to the operation law of the gravity dam, it shows that the selected dam section has the possibility of sliding instability, which is consistent with the actual service situation of the dam. In addition, the hybrid model can be used for the reliability analysis of other structural projects after some improvement and expansion. © 2022 China Water Power Press. All rights reserved.