非高斯随机振动的模拟方法

随机振动台根据功率谱密度（PSD，Power Spectral Density）生成的信号为高斯信号，而实际振动环境有时是非高斯的，因此随机振动实验常常无法准确模拟产品在真实振动环境下的失效情况。通过两个案例分别对均方根值（RMS，Root Mean Square）不随时间变化和随时间变化的非高斯随机振动进行了模拟方法研究。案例1利用Hermite多项式法对高斯信号进行了转换，在保证功率谱密度不变的同时得到了具有指定峭度的RMS不随时间变化的非高斯信号，但该方法对于输入的峭度有限制，当输入峭度大于10时，误差达到了20%。案例2利用一种新方法对实测的RMS随时间变化的非高斯振动进行了模拟，模拟后得到的非高斯信号和实测信号具有相同的功率谱密度、峭度以及概率分布，验证了新方法的准确性。

Simulation of non-Gaussian random vibration

Traditionally, only Gaussian signal can be produced by random vibration controller using power spectral density (PSD). However, actual vibration is usually non-Gaussian, which makes random vibration testing unable to simulate the failure mode that products will experience. Simulation of non-Gaussian random vibration with constant root mean square (RMS) and running RMS was performed respectively using two case studies. Case study 1 uses Hermite polynomial to transform Gaussian signal to non-Gaussian signal. The synthesized non-Gaussian signal has constant RMS, desired kurtosis and the same PSD with original Gaussian signal. Case study 2 uses a new method to simulate measured non-Gaussian random vibration which has running RMS. The synthesized non-Gaussian signal has the same PSD, kurtosis and probability density function with the field data.