An improved method of structural random loading identification

Random loading identification has long been a difficult problem for multi-input-multi-output stationary random vibration problems. In this paper, the inverse pseudo-excitation method is used for dealing with such identification problems, and condition number weighted average method is developed to improve the rank defect of the frequency response function matrix during the process of random loading identification; simulation process is given to compare the result between the presented method and routine method. It proves the optimizing result by use of the method; experiments are designed to certify the feasibility of the algorithm, and achieve good results, and the ill-conditioning problem can be conquered to some extent. The condition number of FRF is used to choose the response measuring point, it reduces the workload, and get a accurate result.