Experimental Study Of The Stationary Random Vibration Load Spectra Identification Method

Loading identification problem is one of the inverse problems in structural mechanics, which is very important for engineering application, The performance to slove this problem didn't carry out for a long time, and many more researches must be done for this problem, especially for the random loading identification. The Inverse Pseudo Excitation Method(IPEM)to identify random vibration, which has been just deduced in theory, is implemented in the finite element program DDJ-W, and so can identify the random forces exerting on the complex structures. This method is easy to perform without less calculation amount. Selecting measured response points by simulation; experiments was performed on a simple structure and a complex one to verify the capability and availability to apply in engineering. The response points, which result is befter than others in simulation, can identify the spectrum of load more accurate, too. The fact illustrates that computer simulation is valuable for engineering application to select the response points to identify loading if the amount of calculation is not too huge. By comparison with the measurement value, the results of experiment show that IPEM is reliable, convenient, efficient and precise, but also has good noise-resistance capability. So it is valuable for engineering applications. This thesis verify that the ill-condition and rank-deficient of the frequency response matrix of the structure is the main factor resulting in the identification errors near the national frequency of the structure. The accuracy of the results was improved by higher measurement precise and using the conventional method for solving pathological equations. The paper discusses finding the frequency segment where the identification errors appeared, so that some measures could be taken on to correct those errors according to conditional coefficient distribution of the frequency response matrix of the structure.