Statistical Identification Of Stochastic Moving Force And Wind Load

Structural health monitoring system has been widely used in modern architecture structure.Long-term monitoring and evaluation can provide important guarantee for the safety and reliability of the served architecture.By the arrangement of sensors in the structure to obtain response information,calculating external dynamic load indirectly has become an important research direction in the field of structural health monitoring,which is valuable reference data of the assessment and design of the structure.At present,most of the force identification research is aimed at deterministic problems.However,in reality,the external force not only changes with time,but also contains random properties,such as vehicle load of bridge-vehicle interaction system,wind load of high-rise structure and so on.To obtain the statistical characteristics of random load can provide more realistic incentive information for structural monitoring,design and further research,and has great value for application.But in the present study,the research on stochastic load inversion problem still has some limitations,and the related research is very limited.Specifically,some scholars need known structure parameters and full observation responses to identify statistic characteristics of random moving load.However,in practical engineering,structure parameters are often unknown and the observed response information is difficult to obtain because of high cost.And under the influence of noise,the interpolation for response order reduction may produce larger deviation.Therefore a method to identify statistical characteristics of random moving load with unknown structural parameters and partial response information need to be put forward.Random wind load inverse identification mainly concentrated in the frequency domain,and research in the time domain is still limited.So it still has great research value.Based on the research background above,the research contents in this thesis can be divided into two parts,random moving load identification of bridge-vehicle interaction system and stochastic wind load identification of high-rise structure respectively.On the basis of related research,stochastic finite element equation of motion can be established to introduce random parameters by a series of decomposition method of stochastic process or random field.Then use our group’s latest research about Kalman Filter with Unknown Inputs to identify statistical characteristics of random force by partial responses.Finally,the numerical simulations validate the method of each part.Main work of this thesis is as follows:In the second chapter,under the condition of the known deterministic structure parameters and the unknown Gaussian uncertainty of moving load,the statistical character of Gaussian random process samples is extracted based on Karhunen-Loeve(KL)expansion,and then establish the new equation of motion which contains the statistical properties.A general method of Kalman filter under unknown inputs(KF-UI)by partial observed response information,which is our group’s research output,can realize statistical character identification of random moving load.In the third chapter the unknown structural parameters has been considered.In reality,the structural parameters are often unknown,such as the mass,stiffness and damping,etc.and the accuracy has much impact on the identification of moving load statistical properties.This chapter put forward a general method of extended Kalman filter under unknown inputs(EKF-UI)to identify the unknown deterministic structural parameters.Then coupled with the method the second chapter propose,the statistical character of Gaussian random moving load can be identify.By these two steps,the identification of structural parameters and statistical character of the moving load can be.realized.The fourth chapter considers the stochastic properties of structural parameters caused by material properties,physical characteristics of structure,boundary conditions and also the uncertainty of components,etc.Therefore this chapter contains not only the stochastic moving forces but also the randomness of structural parameters.Based on the spectral stochastic finite element method,a stochastic finite element equation of motion including both the randomness of the external excitations and internal structural parameters has been established.So by the identification method,more realistic statistical features of structural response and external force,will be obtained.In the fifth chapter,the method of KL expansion to extract the statistical character of random process samples is extend to random wind load identification.Under the known deterministic high-rise structure,the nodes wind load on the structure is assumed about two separated functions of time and space using modal shape as orthogonal basis.The modal superposition method can simplify the wind load whose number is less than observed vibration responses.And based on modal Kalman filtering method under unknown inputs(MKF-UI)which is also our group’s research output,modal force can to identified to inverse the statistical character of random wind load.After theoretical derivation in every chapter,numerical simulation is used to verify the validity of the method.