论文部分内容阅读
金属切削时的振动信号具有独立于尺度和随机的特征,与1/f过程相似。用1/f过程功率谱中的指数γ作为特征量可以用来检测颤振的发生。实验表明,切削颤振的产生伴随着参数γ的急剧的改变。本文提出了一种新的用于估计γ的基于小波的最小二乘算法。与极大似然估计法相比,它具有更广的适用性。通过实验结果对比可以看出,用γ的相对变化量作为特征量更为合适。
The vibration signal during metal cutting has a scale-independent and random feature similar to the 1 / f process. Using the exponential γ in the 1 / f process power spectrum as a feature can be used to detect the occurrence of chatter. Experiments show that the generation of cutting chatter is accompanied by a sharp change of the parameter γ. In this paper, a new wavelet-based least squares algorithm for estimating γ is proposed. Compared with the maximum likelihood estimation method, it has a wider applicability. By comparing the experimental results can be seen, the relative change in γ as a feature amount is more appropriate.