Study On Optimal Placement Of Multidimensional Sensors For Structural Health Monitoring

Optimal sensor placement is one of the key issues of structural health monitoring system, which has an important influence on information collection. Taking into account the actual complex and huge engineering structure, as well as the influence of the field conditions and the construction costs and other factors, only relatively few sensors are decorated on the limited positions. Therefore, a good sensor layout scheme should use less measuring points to get enough structural state information. The previous optimal sensor placement algorithms only consider one direction, and then the multi-dimensional sensors are arranged on these points. In that case, the direction which is considered is optimal, but the other directions are not the most excellent. To solve this problem, this paper proposes a optimal method for multi-dimensional sensors, the contents of the study include the following aspects:(1) Based on the classical triaxial effective independence (EfI3) proposed by Kammer, this paper extends it to the other dimensions and the EfI2&EfI5applicable to high-rise buildings are given respectively. EfI3uses the matrix determinant value to maximize the Fisher information matrix and This paper extends it to other criterions such as the matrix trace to evaluate the Fisher information matrix. Then, the merits and drawbacks of different criterions are compared.(2) To solve the problem that the previous algorithms can not consider the ill-posedness of modal matrix, the optimal method comprehensively considering the coordination of structural response sensitivity and system robustness is proposed, which can make the Fisher information matrix obtain the maximum value by EfI values and ensure that the condition number of modal shape matrix is minimum during the iterative procedure. Finally, taking the Guangzhou new TV tower as an example, the sensor placement configurations that only considering horizontal vibration, and simultaneously considering the horizontal vibration&spatial torsion are compared and analyzed by using three evaluation criteria, i.e., modal assurance criterion, Fisher information matrix determinants, and condition number. Numerical results indicate that the proposed method is efficient and effective.(3) The previous algorithms ignore the redundant information between two degrees of freedom. In order to solve the problem, the rapid expansion method based on redundancy analysis for multi-dimensional sensors is proposed. The last optimal sensor placement program can both ensure the Fisher information matrix obtain the maximum value, while avoid redundant information between the freedoms considered. Finally, taking the Dalian Beida bridge as an example, the sensor placement configurations are compared and analyzed by using three evaluation criteria, i.e., modal assurance criterion, Fisher information matrix determinants, and modal visualization. Numerical results indicate that the proposed method is efficient and effective.(4) Based on mass-weighted effective independence proposed by Kammer, this paper proposes stiffness-weighted effective independence method. In order to improve the efficiency of the iteration and the ability to ensure that the arrangement of the sensor at specific points, and thus the stiffness-weighted effective independence extension method is given, which is extended from the initial configuration results until the desired number of sensors.