Research On Load Identification Based On Fiber Grating Sensors

Load identification belongs to the inverse problem.In the case where the system is known,load can be computed by using system responses.Load identification plays an important role in structural health monitoring,and the methods of load identification which are based on kalman filter and fiber bragg grating sensors are proposed to meet the requirements of structural health monitoring.For the proposed method,Kalman filter is used to suppress noises,and the residual innovation sequence,a priori state estimate,gain matrix and innovation covariance generated by Kalman filter are employed to identify the magnitude and location of load by using a least squares estimator.Kalman filter is famous for its strong stability and higher precision which only need the recent measurement datas and the previous identified values to identify load,so the proposed methods can save computer memory,reduce computational burdens and improve system robustness.The method of load identification based on BP neural network or deep learning is proposed to avoid the complexity of system modeling.The contents are as follows:(1)The method of load identification for linear system is proposed,which is based on Kalman filter.The residual innovation sequence,gain matrix and innovation covariance generated by Kalman filter are employed to identify the magnitude and location of load by using a least squares estimator.The paper establishes the relationship between strain values and nodal freedoms of Bernoulli-Euler beam,so strain values getting from fiber bragg grating sensors can be used as observed values to identify load.Simulations and experiments of Bernoulli-Euler beam systems verify the effectiveness of the algorithm.(2)The method of load identification for nonlinear system is proposed.The method is based on nonlinear Kalman filter and the residual innovation sequence,a priori state estimate,gain matrix and innovation covariance generated by nonlinear Kalman filter are employed to identify the magnitude and location of load by using a least squares nonlinear estimator.Simulations and experiments of nonlinear systems verify the effectiveness of the algorithm.(3)The method of load identification based on BP neural network or deep learning is proposed.First,the method uses responses of system and exact load to train network.Then the network is used to identify load by inputing responses.The proposed method uses the characteristics of BP neural network or deep learning,and can solve the difficulties of establishing model of complex system.Simulations and experiments verify the effectiveness of the algorithm.(4)This paper presents a method to identify load and predict displacement at the same time based on dynamic strain measurement by distributed fiber bragg grating sensor network.In the process of identifying load,displacement can be monitored at the same time which can satisfy the demand for structural control.To prove the effectiveness of the proposed method,numerical simulation and experiment of a beam structure are employed and the results show that the method has good performance.