The Investigation On Parameter Identification Methods For System Of Dymatic Testing Non-stationary Signal

Test signal processing in background of big noise and short samples is an an important topic in the field of Engineering applications, and the signal often involve lower-quality of the test signal data, the data are often modal-intensive, non-linear, the effective sample is short, the noise ratio of data signal is low and the process of non-stationary characteristics. So, in this paper study and realize a new time-frequency domain filter de-noising method and modal parameter identification methods to better solve the engineering related issues. Main contents of this article are as follows:1. To enhance noise ratio of test signal, proposed a linear sweeping signal de-noising method which is based on Fractional Fourier Transform to the Impatient algorithm, as a generalized form of the traditional Fourier transform, there are many traditional Fourier transform features are not available. The algorithm uses the sweep signal of the focusing properties of time-frequency domain, using cover every other operation in the fractional Fourier domain to achieve the purpose of Signal to noise separation,Compared with the traditional filtering method, this method is better and Calculation is smaller.2. In order to better solve test modal parameter identification problem of low accuracy,proposed multi-reference-point Least-squares complex frequency domain method (poly_LCSF), applicable both to the weak damping also applies to strong modal damping-intensive systems. The algorithm uses a linear least-squares-like algorithm is estimated that the overall model coefficients, avoiding the design and optimization of complex nonlinear dependence on the initial value, and successfully applied to experimental data analysis, to achieve better results.3. Based on the above principle of algorithms and other similar methods, using GUI interface tools in MATLAB to develop a simple and practical analysis software. Including the vibration signal pre-processing module, time-frequency processing module and time-frequency domain modal parameter identification module. After verification, the software can be easily applied to voice, and vibrations (such as the flight flutter test sweep signal) a variety of signal analysis.