【摘要】 在铣削加工过程中,为实现在线不停机刀具磨破损的快速检测与预报,提出了一种基于Gabor-GLCM工件表面纹理特征的刀具状态视诊方法。首先采用Gabor滤波器虚部与工件纹理图像卷积并提取Gabor特征;然后在卷积图像上采用GLCM提取二阶统计特征,将提取的特征向量串联,旋转规范化后得到刻画工件纹理的特征集;最后将旋转规范化特征集输入SVM训练分类模型,在实验采集的铣削工件表面纹理图像库中进行了测试。实验结果表明:该方法用时短,分类正确率高达98.667%,提高了刀具磨损状态监测的正确率和实时性。更多还原

【Abstract】 In the process of milling, in order to realize the rapid detection and prediction of tool wear and damage online and no downtime, a tool state visual diagnosis method based on Gabor-GLCM workpiece surface texture features was proposed. Firstly, the Gabor filter imaginary part was convolved with the workpiece texture image and the Gabor feature was extracted. Then the GLCM was used to extract the second-order statistical features on the convolution image. The extracted feature vectors were connected in series and rotation normalization to obtain the feature set of the workpiece texture. Finally, the normalized rotation feature set was input into the SVM to train classification model, and tested in the experimentally acquired milling workpiece surface texture image library. The experimental results show that the method takes a short time and the classification accuracy is up to 98.667%, which improves the correct rate and real-time performance of tool wear state monitoring.更多还原