基于功率谱小波分解的神经网络钻头磨损监测

在钻削过程中，钻削力功率谱与钻头磨损之间具有较强的相关性。被广泛用于钻头磨损监测，但是关于功率谱特征的提取和识别一直没有很好解决．文中采用小波变换对功率谱进行多层分解，提取低频分解系数作为功率谱的包络信息，从而实现对功率谱特征的提取和压缩，并利用BP神经网络对功率谱小波低频分解系数进行融合，实现钻削过程钻头磨损状态的智能识别．试验结果表明：该方法可有效实现功率谱特征提取，经训练的神经网络具有较高的识别精度和推广能力．

Artificial Neural Network Based Drill Wear Monitoring Using the Wavelet Decomposition of a Power Spectrum

In the drilling process, the power spectrum of a drilling force is closely related to the drill wear. This relationship is widely applied in the monitoring of drill wear. But the problem of how to extract and identify the features of power spectrum have not been completely sloved. This paper achieves this through the multilayer decomposition of the power spectrum by using the wavelet transform and the extract of the low frequency decomposition coefficient as the envelope information of the power spectrum. Intelligent identification of the state of drill wear is achieved in the drilling process through fusing the wavelet decomposition coefficients of the power spectrum by using BP neural network. The experimental results show that the features of power spectrum can be extracted efficiently through this method, and the trained neural networks have high identification precision and the ability of extension.