论文部分内容阅读
为了提高对切削加工过程中颤振的识别能力,提出一种结合模态分解和支持向量机的分类方法,对颤振信号进行自动识别。首先利用经验模态分解法把颤振信号分解成若干个本征模式函数分量,去除原始信号相关性,突出模式函数分量的主特征,构建出特征向量并进行归一化处理,之后,把特征向量输入SVM模型,判断颤振是否发生。分别采用神经网络模型、PCA-SVM模型、EMD和SVM模型对特征向量进行学习与识别,EMD和SVM模型识别率达到95%,优于前两种模型。实验结果表明,该方法能够有效地识别切削加工过程中的颤振。
In order to improve the ability of chatter vibration detection in cutting process, a classification method based on modal decomposition and support vector machine is proposed to automatically identify flutter signals. First, the empirical mode decomposition method is used to decompose the flutter signal into a number of eigenmode functional components, remove the correlation of the original signal, highlight the main features of the mode function components, and construct the eigenvectors and normalize them. Then, Vector input SVM model to determine whether chatter occurs. The neural network model, the PCA-SVM model, the EMD model and the SVM model are respectively used to learn and recognize the feature vectors. The recognition rate of the EMD and SVM models is 95%, which is better than the former two models. The experimental results show that this method can effectively identify the flutter in the cutting process.