Interval Inversion Analysis Of Structural Damage Under Uncertainties

Various damages may occur in existing structures and accumulate over time,under the influence of natural environment and human factors.The damages will reduce structure performance and cause safety accidents,so it is very meaningful to develop effective methods for damage detection and monitor the structure health status in real time to ensure the safe operation over the structure.Structural damage identification method based on deterministic framework has laid a solid foundation for successful health diagnosis and safety evaluation.But due to the diversity of construction,limitations of measurement environment,loads of long-term and other unexpected events,there still exist many uncertainties in practical engineering.The damage identification method does not directly consider the influence of uncertainty factors,nor have the ability to solve the uncertainty problems.The inversion result is not good when used in the actual structure.Therefore,it is in urgent need to develop the damage identification method with uncertainty for effective damage detection.In this paper,two inverse methods are proposed to efficiently identify structural damage,in which circumstances uncertainties in measured responses and model are considered.The following studies are carried out in this dissertation:(1)The uncertain responses in observation are treated as Guass noise,while that in modeling regarded as structural parameters,described by intervals within specific bounds.An interval inverse method that combines the extended kalman filter and genetic algorithm is proposed.First,the optimal estimate of state vector in extended kalman filtering is set as objective function.The genetic algorithm is then applied to search for the upper and lower bounds of the state vector within the scope of the uncertain parameters.Thus the interval inversion of uncertain parameters is achieved.(2)The uncertain responses are treated as unknown-but-bounded noise.Based on the set membership filter,an interval inverse method is proposed.First,the state vector of structure is treated as ellipsoid set.Then,the nonlinear systems in the process of inversion are linearized,with the current estimation and the interval mathematics to acquire the bounds.Next,the bound of the linearization error is combined together with linearization error or system error to acquire a new disturbances ellipsoid,which can be seen as system noise.The state ellipsoid can be estimated recursively by applying a linear set-membership estimation algorithm at each time step.Finally,the damage identification of structure with uncertainty can be realized,through taking the boundary of ellipsoid as interval value of parameter identification and the central value as the best estimate.(3)This paper applies regularization technique to solve the ill-condition of inverse problem.The regularization of optimal estimate in set membership filtering can be converted into inequality constraints,which can then get the state vector ellipsoid updated.A priori information constraints,as a hyperplane,could further update the state vector ellipsoid.Both measures can improve the precision of the inversion algorithm.