【摘要】 文章使用支持向量回归方法(SVR)成功地对切削颤振状态趋势进行了预测,同时提出了一种新的信号特征提取方法。首先逐次对切削信号进行小波包分解,然后计算各频带区间内的能量并对能量进行归一化处理,于是得到了信号在各区间的能量分布图以及各区间的能量变化曲线,从能量的变化曲线可以很清楚的看到,各区间的能量分布很好地反应了切削颤振过渡的特征。最后通过SVR算法对能量变化趋势进行回归预测,与实际曲线进行比较,预测结果基本能反应出能量的变化趋势,从而为切削颤振的预报奠定了较好的基础。更多还原

【Abstract】 This article adopted support vector regression algorithm succeeded in predicting the cutting chatter trend,at the same time,a new characteristic extracting method was delivered.firstly,bootstrapping analysing of the cutting signal using wavelet package decomposition,then computing the energe in each spectrum section and scaling,now getting the energy distribution chart and transition curve of different section,from the curve we could clearly see that it was really a very good representation of the cutting chatter transition characteristic.At last,using SVR algorithm to predict the trend of the curve,comparing it to the real curve,the result basically represent the transition trend.Consequently,it established a fine fundation for cutting chatter forecast.更多还原