Optimal Multi-type Sensor Placement And Identification Of System

The sensing system and system identification algorithm play important role in structural health monitoring(SHM)system.Due to practical limitations,it is often difficult to place a great number of sensors in every key location.Therefore,for the purpose of obtaining enough structural information under limited observations,it is critical to develop the optimal sensor placement techniques.Different types of sensors are used in practical structural health monitoring system for assess the local damage and overall status of the structure.For most existing researches,the simultaneous optimization of multiple types of sensors is a challenging process.Moreover,many sensor placement methods are performed on the premise that the external excitation is assumed to be known.However,in many practical situations,it is not easy to obtain the input information.As known,the extended Kalman filter(EKF)provides a promising way for the identification of structural parameters in the time domain.It is critical to develop the EKF algorithm under unknown excitation and limited observation conditions.In the process of structural damage development,nonlinear behavior will occur.Identifying nonlinear restoring force is an important index to evaluate the degree of structural damage.It is inefficient to characterize the nonlinear restoring force using analytical models.Therefore,the idea of model-free restoring force identification provides a promising way for solving such problem.To circumvent the shortcomings mentioned above,based on Kalman filer,the approaches for optimal sensors placement,parameters identification and model-free nonlinear restoring force identification are proposed and described in this thesis.The main contents are listed as follows:(1)A multi-type sensor optimal placement(OSP)algorithm under unknown excitation is proposed.The sensors are selected via the optimization process by minimizing the estimation errors of the reconstructed multi-scale responses.Three types of sensors including accelerometers,displacement and strain measurement sensors are considered.The unknown inputs applied to the target structure are estimated at the same time.The algorithm based on the modal Kalman filter with unknown excitation.Three numerical examples are given to verify the effectiveness of the proposed approach.Results show that the proposed OSP approach is capable of deploying the multi-type sensors at their optimal locations with the aim of accurate reconstructed responses being obtained.(2)By using the elementary transformation matrix,an EKF-based approach is proposed to identify structural state under unknown inputs.The effectiveness of the proposed method is verified by three linear examples and frame structures equipped with two different nonlinear models.(3)A model-free nonlinear restoring force identification approach is proposed based on EKF method mentioned above.By considering the nonlinear restoring force as an unknown "pseudo-external excitation",the aforementioned approach employed for model-free identification.In order to avoid low-frequency drift caused by the accumulation of numerical integration errors,data fusion technology is combined.A frame equipped with different types and numbers of nonlinear components is analyzed as numerical example.The results indicated that the proposed method can effectively identify structural parameters and nonlinear restoring forces.