Control Algorithms And Some Issues For MIMO Random Vibration Test

MIMO random vibration test can simulate the real vibration environment better than SISOrandom vibration test, it can find more potential reliability and security problems, and has importantmeaning for the development of economic and national defense. MIMO random vibration test hasbeen required in USA MIL-STD-810G, more attentions has been paid on MIMO random vibrationtest. Based on control theory for MIMO linear vibration test, related issues about MIMO randomvibration test control will be studied in this dissertation.Base on the all phase spectral estimation theory of digital signal process, all phase frequencyresponse function estimation method is proposed, which can improve the control results of MIMOrandom vibration test. The affect to MIMO random vibration test caused by the inaccurateness offrequency response function estimation is analyzed. Theory of all phase spectral estimation theory isintroduced and is applied into frequency response function estimation, and all phase frequencyresponse function estimation method is formed. This estimation method improved the control resultsof MIMO random vibration test.The relationship between memory excitation signals and measured response signals and thereason of their asynchronism are analyzed, a double correlation method to synchronize these twosignals is proposed. After the introduction of working process of VXI digital signal transmission andacquisition instrument, the synchronization problem between sending signals and sampling signals ispresented. Base on the relationship between exciting signals and response signals, a doublecorrelation method is proposed, in which the correlation function calculation and random decrementmethod are used. Double correlation method can get the time history relationship between excitingsignals in computer memory and response signals measured by VXI.The "window" and "overlap" process of the time domain randomization process are analyzed; animproved generation method of random exciting signal is proposed. It is found in tests that thefrequency spectrums of generated signals are not the same with the expected drive spectrums, whichwill result in the over range spectrum in control result of MIMO random vibration test. After theanalyses of "window" and "overlapping" of time domain randomization process, a four frameoverlapping method is proposed. Under the circumstance of not changing the number of spectrumlines, this method improves the exciting signals and the random test control results.Factors that affect the control accuracy of MIMO random vibration test are analyzed; PID controlalgorithm with variable arguments is proposed to make rapid correction and to improve control accuracy. The idea of PID control is applied into MIMO random vibration test control. Theproportional control, integral control and differential control is introduced in test control under thelogarithmic coordinates, and then been transformed into forms under linear coordinates. The valuesof arguments of PID control are adjusted according to the errors between response spectrums andreference spectrums. The PID control algorithm can make the rapid correction of response spectrums,and also improve the random vibration test control accuracy.Compare analsis among control algorithms are made, Matrix power control algorithm (MPCA) isproposed, which can steadily and convergently control the auto spectrum and cross spectrum. Thetheory of matrix power control algorithm is deduced. A compare analysis among difference controlalgorithm (DCA), proportional root mean square control algorithm (PRMSCA) and MPCA is made.The error assumption and error correction of DCA is based on difference operation, while the errorassumption, error calculation and error correction is base on matrix multiplication operation. Thisdifference forms is why MPCA can avoid the "negative value problems". At last simulations andtests are made, the results proved the auto spectrum and cross spectrum control ability of MPCA.