Let's discuss the most commonly used excitation techniques for modal analysis today. These are random, burst random, sine chirp and digital stepped sine. But before we discuss the excitation techniques themselves, there are a few basics that we need to discuss first. Let's try to categorize the different techniques and explain when to use which technique. First of all, let's break up the excitations into deterministic and non-deterministic (or random) excitations.

让我们讨论一下当前模态分析最常用的激励技术。这些激励技术是随机、猝发随机、正弦扫频和数字步进正弦。但是在我们讨论这些激励技术之前，需要先讨论点基础知识。我们设法对不同的技术进行分类，并解释何时使用何种技术。首先，我们将激励分成确定性激励和不确定性（或随机）激励。

 

Deterministic signals are those that can be described at any point in time by a mathematical function - they can be determined. Typical signals of this type are sinusoidal in nature, such as sine chirp and digital stepped sine. Random signals, on the other hand, can not be described by a mathematical function but are rather described by their statistical characteristics. Typical signals of this type are random and burst random.

确定性信号是指在任意时刻上，可以用数学函数来描述的那类信号 — 可以确定它们。这类典型信号实际上是正弦的，例如正弦扫频信号和数字步进正弦信号。另一方面，随机信号不能用数学函数来描述，而是更倾向于用它们的统计特性来描述。这类典型信号有随机信号和猝发随机信号。

In general, we use deterministic signals on linear systems. We also use deterministic signals to determine if a system is linear by performing a linearity check. We use random signals to average slight nonlinearities in a system due to things such as rattles. If we have a structure that has gross nonlinearities, then we need to stop and think just how useful the results of a linear modal analysis will be. But understanding the difference between these two categories helps in deciding which technique will provide the best measurement. Depending on the system being tested, you may want to document the linearity of the system under test, or you may want to linearize any slight nonlinearities that exist.

一般情况下，我们对线性系统使用确定性信号。也利用确定性信号，通过进行线性度检查来判断一个结构是否是线性的。我们利用随机信号来平均掉因某些情况，如震颤，引起的系统轻微非线性。如果结构具有严重的非线性，则我们需要停下来，考虑一下线性模态分析的结果在多大程度上有用。但是搞清楚这两类信号的差别，有助于确定何种激励技术可以得到最优的测量结果。你或许希望记录被测系统的线性，或者，你可能想对实际上存在的少许轻微非线性进行线性化处理。

Now first, let's consider a random excitation. Random is used quite widely for general vibration testing today. But it is not considered one of the best techniques for acquiring FRF measurements for modal testing (although it is still often used). The random nature of the signal excites the structure with varying amplitude and phase as averages are collected. This tends to average any slight nonlinearities that may exist in the structure. While this is a benefit, the signal never satisfies the periodicity requirement of the FFT process. Therefore, leakage is a tremendous problem. Even with a Hanning window applied, the resulting FRFs will always suffer from leakage; the peak amplitude will be affected and there will be an appearance of more damping in the structure due to the leakage and windowing effects. A typical measurement sequence is shown in Figure 1. The resulting FRF and COH are shown in Figure 2. Notice how the coherence dips at the resonances of the system; this is a characteristic of random excitation.

现在，我们首先考虑随机激励。当前对于一般的振动测试，随机信号应用的很广泛。但是，对于模态试验中进行FRF测量结果的采集来讲，没有人认为它是一种最优的技术（尽管它依然常被使用）。信号的随机特性对结构进行激励，幅值和相位不断变化，同时采集到平均结果。这往往会平均掉结构上可能存在的少许轻微非线性。尽管这是个优点，但信号完全不满足FFT处理的周期性要求。因此，泄漏是个大问题。即使施加了汉宁窗，所得的FRFs也总会受到泄漏的不良影响；会影响峰的幅值，并且结构上的阻尼看起来变大了，这是泄漏和加窗效应引起来的。典型的测量顺序如图1所示。所得到的FRF和COH显示在图2中。注意相干在系统共振频率附近是怎样下降的。这是随机激励的特性。

Now, let's consider a burst random excitation. The only difference is that the random signal is only used during a portion of the data capture. If a pretrigger delay is also used, then the signal is totally observed within one sample interval. Therefore, the signal satisfies the periodicity requirement of the FFT process. This means that no leakage will occur and no window is needed. Of course, both the input and response signals need to satisfy this requirement. This is easily done for most structures.  This signal is well suited for averaging out slight nonlinearities that may be found in the measurement.  A typical time measurement is shown in Figure 3. Notice that the excitation is terminated such that the response signal also decays to zero within the sample interval. The resulting FRF and COH are shown in Figure 4. Notice the improvement in the measurement and coherence when compared to Figure 2.  The peaks are much sharper and better defined; the coherence is especially good at the resonances.

现在，我们来考虑猝发随机激励。仅有的差别是，只在数据采集过程的一部分时间内使用随机信号。如果同时利用预触发延迟，那么在一个采集时间段内，可以观测到完整的信号。因此，信号满足FFT处理的周期性要求。这意味着不会产生泄漏，无需加窗。当然，输入和响应信号二者都需要满足这个要求。对大多数结构，这点易于满足。这个信号非常适合于平均掉测量结果中可能存在的轻微非线性。一个典型的时间测量结果显示在图3中。注意到，在采集时间范围内，激励中断以致响应信号也衰减到零。所得的FRF和COH显示在图4中。与图2相比较时，可以注意到测量结果和相干的改善。谱峰更尖锐、更明确；在共振频率附近，相干格外地好。

Now, the sine chirp is a fast sweep from low to high frequency within one sample interval of the analyzer. The signal repeats and therefore satisfies the periodicity requirement of the FFT process.  This means that no leakage will occur and no window is needed. Of course, the signal must be played continuously so that the structure steady state response is achieved. The resulting FRF and COH are shown in Figure 5. The measurement is very similar to the results from the burst random test. By changing the input force level applied to the system, linearity checks can be easily made using this excitation technique.

目前，正弦扫频是在信号分析仪的一个采样时间范围内，从低频到高频进行快速扫频。信号重复发生，因此满足FFT处理的周期性要求。这意味着没有泄漏发生，无需加窗。当然，信号必须连续发生，这样可以得到结构的稳态响应。所得的FRF和COH显示在图5中。这个测量结果跟猝发随机试验得到的结果非常类似。利用这个激励技术，通过改变施加到系统的输入力的幅度，很容易进行线性度检查。

Finally, the digital stepped sine technique requires that a single frequency, coincident with an analyzer spectral line, is used to excite the system. Since the signal is guaranteed to be periodic with regards to the FFT process, no leakage occurs and no windows are necessary. Since it is not broadband in nature, this technique is the slowest of all techniques because each spectral line is evaluated individually. However, it is excellent for documenting nonlinearities and is likely to produce the best measurement of all the excitation techniques above.

最后，数字步进正弦技术要求如下，用一个单频的，与信号分析仪谱线相一致的信号来对结构进行激励。由于可以保证信号对于FFT处理是周期的，因此没有泄漏发生，也无需加窗。因为本质上说,它不是宽带的，因为需要单独求取每根谱线的数值，因此这项技术是所有技术中最慢的。但是，它对于记录非线性是非常好的；并且在上述所有激励技术中，它有可能得到最优的测量结果。

 

When comparing the techniques, the burst random and sine chirp will produce similar results if the system is linear. In general, the random measurement will always suffer from leakage and the quality of the measurement will suffer when using this technique. To illustrate the degradation of the measurement when using random excitation, Figure 6 compares random and burst random with an expanded look around the first resonant peak of the system. The random signal contains alot of variance and the peak is distorted at resonance (where the coherence is known to dip). In fact, there almost appears to be two modes at that frequency; this is due to the distortion of leakage. The burst random measurement is clean and sharp. Clearly, the burst random measurement is the better of the two measurements.

对这些技术进行比较，如果系统是线性的，猝发随机和正弦扫频可以产生类似的结果。一般来讲，当使用随机激励时，随机测量结果总是会受到泄漏的不良影响，测量质量会变糟糕。为了说明使用随机激励时测量结果的糟糕状况，根据系统第1阶共振峰附近的放大图形，图6比较了随机和猝发随机技术。随机信号方差很大，且在此共振频率附近，谱峰失真（可知此处相干变小）。实际上在那个频率附近，看上去似乎是有两阶模态。显然，在这两个测量结果中，猝发随机测量结果更好些。

We could spend alot more time discussing all the details of each of the techniques (as well as others not mentioned) but there isn't enough time right now to cover everything. Maybe another time we can discuss each of the techniques in more detail. But this quick overview should give you what you need to know. If you have any more questions on modal analysis, just ask me.

我们本可以花更多的时间来讨论每种技术的所有细节（以及其他没有提到的技术），但是当下没有足够的时间来面面俱到。下一次我们或许可以更详尽地讨论每种技术。但是这个概述应该是讲到了你所需要知道的内容。如果你有模态分析方面的其它任何问题，问我好了。