The main excitation technique used in modal testing today is burst random, used more often than others such as random, sine chirp and digital stepped sine. These are discussed with some brief comments. (Other techniques such as pseudo-random, periodic random, burst chirp and others are variations on these signals and are not discussed at length here.)
Random excitation was one of the first excitation techniques used because it was simple to create. The problem with random excitation is that the signal is never periodic in the sample interval of the FFT measurement and requires a window (commonly a Hanning window) to mediate the effects of leakage. Unfortunately, even with a window applied, the frequency response measurement suffers from leakage especially at the resonant peaks of the measurement.
图9 – 随机激励的典型测量步骤
图10 – 输入，输出随机时域信号，以及FRF和相干
Sine chirp is an excellent technique for testing systems that are fairly linear. This signal is a very fast swept sine where the frequency is swept from low frequency to high frequency within the time of one sample of the FFT analyzer. As a result, the signal is periodic in the sample interval once steady state response is achieved. This signal does not require any windows and does not suffer from leakage.
图11 – 正弦扫频激励的典型测量步骤
图 12 – 输入和输出正弦扫频时域信号，以及FRF和相干
Burst random excitation was developed in the early 80s and has remained as one of the more commonly used excitation techniques for experimental modal testing. The vast majority of all shaker modal tests today employ burst random technique.
Burst random is formed as follows: A random excitation is generated but is only applied for a portion of the data block. In this way, the excitation signal is totally observable within one sample interval of the FFT analyzer and there is no need for the use of windows since there is no leakage associated with the captured signal. (Note: Providing that the response measured on the structure is also totally observable within one sample interval of the FFT analyzer then there is no need for the use of windows since there is no leakage associated with the captured signal.) However, once the excitation is turned off, the structural response will die exponentially depending on the damping associated with the structure. If the response of the structure does not die out within one sample interval, then the burst should be shortened such that the response does end before the end of the sample interval. The burst can be controlled by specifying the percentage of the block over which the excitation is to be applied. Generally, this can be accomplished with most structures.
图13 – 猝发随机激励的典型测量步骤
图14 – 输入和输出猝发随机时域信号，以及频响和相干