Research On SEA Parameters Identification Based On FERA

Statistical Energy Analysis (SEA) is an effective way to solve the high-frequency dynamics response prediction, which is widely used in the domains ofaviation and aerospace, the vehicle, the navy and civil engineering. Theevaluating reliability of the SEA method depends on that whether the statisticalenergy analysis parameters can be estimated correctly. To solve this problem, inthe present paper a new method of OKID/FERA was conduct, which iscombining the Fast Eigensystem Realization Algorithm (FERA), OKID algorithmwith Wavelet Packet Denoising method.Firstly, stable power flow model with stable state and state space modelbased on the balance equation of the transient power flow were given. Andparameters identification method with stable state statistical energy analysis andquasi-steady state statistical energy analysis were introduced, they are powerinput method (PIM) and eigensystem realization algorithm (ERA). Then,OKID/FERA identification algorithm of the quasi-steady state statistical energyanalysis was developed, where Markov parameter was gained by using the OKIDalgorithm, and the measuring signal was decomposed and reconstructed byWavelet Packet Denoising method to denoise. Adopting identification to get theeigenvalues and eigenvectors, and combining with optimization idea of constraintoptimization problems to gain the modified formula of the SEA, thus the initialstatistical energy analysis model was modified.Secondly, validating the rapidity and accuracy of OKID/FERA through aseries of numerical simulation, where FERA and ERA were compared to identifydifferent dimension matrix and different number subsystem models. By thecomparisons of OKID/FERA simulation results between the denoise before andafter, to identify two-subsystem model parameters and model updating results,thus the correctness of the OKID/FERA was validated. And it was proved thatWavelet Packet Denoising method is available. Similar simulation applied to thestructure of macro-satellite, the satisfactory results were gained.Finally, using OKID/FERA to identify SEA parameter of the direct bendingL-shaped plate, the actual input power and the response energy were obtained from the experimental measurement data. Identified model of the L-shaped platewas gained by OKID/FERA. Then, utilizing SEA model improvement, thestatistical energy analysis parameters were obtained. Comparing with powerinput method, a good agreement was gained. Thus it was indicated that theexperimental result further verify the validation of the algorithm.