基于谐波小波包和改进功率谱的球磨机振动信号特征提取

Feature Extraction for Ball Mill's Vibration Signal Based on Harmonic Wavelet Packet and Improved Power Spectrum

为有效提取球磨机振动信号特征实现负荷识别,提出一种基于谐波小波包和改进功率谱相结合的特征提取方法。首先应用谐波小波包分解将振动信号分解到不同频段,对各频段的小波系数进行改进自相关得到自相关函数,再通过自相关函数加2阶Hanning自卷积窗后进行离散傅里叶变换求取各频段功率谱,最后引入频偏结合能量重心法对功率谱进行校正,获得精准的各频段层最大值对应频率,据此获取球磨机内部负荷的特征向量。基于该方法应用于工业球磨机实测的结果表明,文中提出方法提取的特征量具有更好的区分度和更高的正确识别率,可作为负荷识别模型的输入实现球磨机负荷的实时检测。

In order to extract ball mill's vibration signal features to realize load identification,a feature extraction method based on harmonic wavelet packet and improved power spectrum was proposed.Firstly,based on harmonic wavelet packet decom⁃position,the vibration signal was decomposed into components in different frequency bands.Then,the autocorrelation function of the component signals was computed by the improved autocorrelation.By constructing the second⁃order Hanning self⁃convolution window function,discrete Fourier transform(DFT)was performed on the autocorrelation functions to obtain the power spectrum.Fi⁃nally,spectrum correction which based on energy barycenter method and frequency deviation method was used to obtain the pre⁃cise frequency of the power spectrum maximums in every layer.The frequency was used as features vector to reflect the inner load of ball mill.The test results based on real industrial ball mill operation data show that the features extracted by the proposed meth⁃od have a better differentiation degree and a higher correct recognition rate.It can be used as the input of load identification model to realize the real⁃time detection of ball mill load.