基于经验模式分解和BP神经网络的刀具状态监测方法

为提高刀具检测识别率,采用经验模式分解法分解不同状态刀具的切削声音信号,计算各信号分量的能量百分比,形成特征向量。利用三次样条插值方法对特征向量进行插值,插值后的特征向量保留了原始信号的所有特征成分,应用BP神经网络识别刀具的状态。实验结果表明:经验模式分解方法和BP神经网络相结合可以有效识别铣刀状态,平均识别率达86%。

Approach to monitoring tool conditions based on empirical mode decomposition and BP neural network

This paper proposes a new monitoring approach capable of improving the detection rate ofcutting tool condition. This approach functions by decomposing the cutting sound signals of different tool states using empirical mode decomposition method, calculating the percentage of the energy of each signalcomponents to form a feature vector, interpolating the feature vector using cubic spline interpolation method, retaining all the characteristics of the original signal in the interpolated feature vector, and thereby achieving cutting tool state recognition using the BP neural network. The results demonstrate that the combination of the empirical mode decomposition method and BP neural network enables an effective identification of the milling cutter states, with the average recognition rate up to 86%.