Study On Optimal Sensor Placement And Damage Identification In Structural Health Monitoring

Large-Scale building structure encountered natural or Man-Made disasters,many weak links will cause a huge security risk.In order to understand the dynamic healthy situation of structures in the harsh environment and provide security warning of potential risk in real time,the structural health monitoring technology based on dynamic parameters of infrastructures is widely used today.The structural health monitoring based on dynamic parameters of infrastructures can provide warning information of potential damage of buildings.Based on warning information,the engineers can make quick,stable,elaborate,and effective plan to ensure the safe operation of the structure during construction and service.In this thesis,a series of researches have been conducted on the problems of structural health monitoring based on structural dynamics parameters.In the beginning,a microhabitat frog leaping algorithm(MFLA)is proposed to solve the optimal acceleration sensor placement of Large-Scale buildings.Then,the AMUSE damage identification method based on structural vibration mode is proposed in this thesis to reduce the noise impact in damage identification.Atlast,in order tosolve the problems of the real time damage identification of Multiple-Level building structures with environmental excitation under noise conditions,a damage identification method of Multiple-Level structural chain model based on displacement reconstruction filter and random disturbance load is proposed.The main contents and conclusions are as follows:(1)The optimal acceleration sensor placement of Large-Scale buildings is studied in this thesis.A microhabitat frog leaping algorithm is presented to solve these kinds of problems.Some influence factors(anti noise ability,linear independent degree,calculation precision and redundancy rate)are taken into account by the objective function.In order to improve the calculation precision and efficiency,the reversed genome designed generating frog process,microhabitat technology process,variable step searching method etc are proposed.Then,the method proposed in this part is applied and compared with other popular used methods using an engineering case.The results show that the calculation precision and efficiency are improved by our method,which is applied for solving such kinds of problems.(2)The AMUSE damage identification method based on structural vibration mode is proposed in this thesis to reduce the noise impact in damage identification.The process of AMUSE algorithm is introduced in detail,and the method is simplified and proved by mathematics.The concept of "reversible modal wave" is proposed,and the modal information of the structure is transformed into a reversible modal wave.The vibration mode information and noise information of the structure are regarded as the signal sources of independent random variable,and the singular value decomposition blind source separation is carried out.The mathematical derivation and theoretical proof of the key parts are given.Based on the engineering examples,the extremely complex conditions of monitoring are simulated.No noise condition,noise condition,traditional EMSDI method,AMSE damage identification method proposed in this thesis is tested respectively.The results show that expected damage identification results cannot be obtained by the traditional EMSDI method in the noisy environment.The AMUSE damage identification method proposed in this thesis can meet the basic damage identification requirements in the noisy environment.These works provides a feasible way for the damage identification problems under harsh monitoring conditions.(3)Based on the real time damage identification of Multiple-Level building structures with environmental excitation under noise conditions,a damage identification method of Multiple-Level structural chain model based on displacement reconstruction filter and random disturbance load is proposed in this thesis.The collected acceleration data is reconstructed into the displacement data and velocity data through the displacement reconstruction technique and numerical differentiation.The noise is eliminated after the data conversion.Then,the displacement and velocity data is put into the process of damage identification without acceleration data.Combined with the Multi-Degree-Freedom chain model and stochastic load theory,the theory of damage identification is developed and perfected.The influence of external excitation force is eliminated in the dynamic equation through the mathematical properties of the cross correlation function and white noise load.Then,we developed the damage index to identify the location of damage.In our work,the damage state of structure can be judged by the response of Multiple-Level building load by external random disturbance environment load.Moreover,the use of building is not disturbed by the monitoring works.At the same time,the external noise interference can be excluded greatly by this method,so the anti noise performance and identification accuracy is improved.(4)The damage identification method of Multiple-Level structural chain model based on displacement reconstruction filter and random disturbance load is applied in some cases in this part.The effect of displacement reconstruction,the process of damage identification,the accuracy of damage identification,and the anti noise ability are discussed.The displacement reconstruction effect and anti noise ability of the simple sine acceleration time history,the displacement reconstruction effect and anti noise ability of acceleration time history of Single-Degree-Of-Freedom system under viscous damping by simple harmonic forced vibration are discussed.Then,A 12-Story steel frame model is taken as an example.The damage identification method of Multiple-Level structural chain model based on displacement reconstruction filter and random disturbance load is tested.The damage identification of the structure without damage,the damage identification of the structure with damage,the damage identification without displacement reconstruction filter under noise condition,the damage identification with displacement reconstruction filter under noise condition are tested respectively.The results show that the damage location of the Multiple-Level structure can still be identified in the noise condition by the method proposed in this thesis.Compared with the traditional ways,the damage identification accuracy and anti noise performance are improved in our work.The method proposed in this thesis is verified suitable for the real time structure health monitoring of Multiple-Level building structures under external random disturbance environment load.