Structural Damage Assessment And Reliability Analysis Based On Modal Interval Algorithms

It is of great significance to consider the uncertainties in the assessment of structural damage,safety and remnant life.When dealing with uncertainties,stochastic or fuzzy theory requires adequate data or expert experience,which is often infeasible in real-world engineering problems.On the contrary,interval theory needs much less data to determine the upper and lower bounds of a parameter,showing better applicability.Unfortunately,the phenomenon of interval overestimation always appears for traditional interval algorithms,which highly restrain their applications to practical problems.Due to it,this thesis adopts the theory of modal interval analysis that can avoid the drawback of interval overestimation.Combined with the modal interval analysis,an improved analytical redundancy reduction method has been proposed to assess the damage of structures having uncertain parameters.Meanwhile,multi-objective optimization algorithms were used to identify the damage severity.Moreover,structural reliability can be calculated using the improved modal interval algorithm and then the remaining life of a structure is evaluated.The main research contents and achievements are listed below.Firstly,an improved analytical redundancy reduction method for vibration-based damage detection was proposed.Then via combining the modal interval algorithm,a damage assessment method was developed for structures having uncertain parameters.Secondly,a deterministic optimization method using mechanical constraints as objective functions has been given,which can identify the damage location and severity.At this moment,in order to avoid interval overestimation,a cyclic interval inverse model,together with a two-layer nesting interval inverse model based on the genetic optimization algorithm,were also proposed for accurately assess the damage severity of a structure.The former model is relatively time-consuming,while the latter model is more cost efficient showing a better capability for real-world problems.Thirdly,modal interval algorithms were adopted to calculate the non-probabilistic reliability of a structure.And a non-probabilistic&fuzz system reliability design method was developed,as well as a hybrid model containing probabilistic,fuzzy and non-probabilistic reliability.The design method can obtain precise fuzzy membership functions and the hybrid model may consider three different uncertain factors.The uncertainties of a full-hall framework support system were taken into account as the interval variables aiming to investigate the non-probabilistic reliability of the system.The result can be regarded as a supplement to probabilistic reliability.Meanwhile,the structural damage prognosis based on non-probabilistic reliability was performed and the residual life of a steel beam was predicted using its historical data.Lastly,the feasibility and reliability of the proposed methods have been verified against different numerical and experimental examples.