Study on the Technology of Tool Wear Monitoring by Modifying Least Square Support Vector Machine via Kalman Filter
-
摘要: 针对刀具磨损状态先验样本少和常规神经网络识别模型收敛速度慢、易陷入局部极小值等问题,提出了基于最小二乘支持向量机(LS-SVM)的刀具磨损量识别技术,并针对模型输出存在系统误差而降低刀具磨损量识别精度的问题,引入卡尔曼滤波算法对时序监测结果进行修正,实现小样本下的刀具磨损量的精确识别。以车削加工为研究对象,采集加工过程中的切削力信号,应用小波包分析技术提取反映刀具磨损状态的特征信息,作为LS-SVM的输入样本,并对模型进行学习训练,完成对刀具磨损状态的识别,最后采用卡尔曼滤波修正其时序监测结果。实验结果表明:LS-SVM模型能高效地实现刀具磨损量识别,需样本数较少,训练速度快,通过卡尔曼滤波修正后的磨损量识别结果精度更高。Abstract: Tool wear state directly affects the product quality, productivity and cost. The tool condition monitoring system is conducive to tool preventive maintenance. The prior samples for monitoring model are limited, and the conventional neural networks recognition model has some drawbacks such running into local minimum value easily, slow convergence rate and so on. In view of these situations, a tool wear monitoring method based on Least Squares Support Vector Machine (LS-SVM) was proposed. Aiming at the systematic error existing in tool wear monitoring result to affect the precision of LS-SVM model, Kalman filters algorithm was proposed to modify the monitoring result. The cutting force signals were measured as monitoring signals. Features extracted by wavelet package transforms as model inputs. Tool wear can be got by the corrected and trained LS-SVM model. The experimental result shows that the LS-SVM model can efficiently carry out tool wear monitoring. Althrough the BP neural networks has better precision, faster learning ability and less training samples. The tool wear monitoring results modified by Kalman filter are more close to the real wear.
-
Key words:
- algorithms /
- backpropagation algorithms /
- convergence of numerical methods /
- efficiency
-
[1] Vallejo Jr A G, Nolazco-Flores J A, Rubén Morales- Menéndez, et al. Tool-wear monitoring based on continuous hidden markov models[J]. Lecture Notes in Computer Science,2005,3773:880-890 [2] Li W L, Fu P, Cao W Q. Tool wear states recognition based on frequency-band energy analysis and fuzzy clustering[J]. Proceeding of 2010 Third International Workshop on Advanced Computational Intelligence,2010,12(57):162-167 [3] Teti R, Jemielniak K, Donnell G O', et al. Advanced monitoring of machining operations[J]. CIRP Annals-Manufacturing Technology,2010,59(2):717-739 [4] Roth J T, Djurdjanovic D, Yang X, et al. Quality and inspection of machining operations: tool condition monitoring[J]. Journal of Manufacturing Science and Engineering,2010,132(4):1-16 [5] Abellan-Nebots V J, Fernando R S. A review of machining monitoring systems based on artificial intelligence process models[J]. The International Journal of Advanced Manufacturing Technology,2010,47(1-4):237-257 [6] Dimla E, Snr D. Sensor signals for tool-wear monitoring in metal cutting operations-a review of methods[J]. International Journal of Machine Tools and Manufacture,2000,40(8):1073-1098 [7] 王国锋,李启铭, 秦旭达,等.支持向量机在刀具磨损多状态监测中的应用[J].天津大学学报,2011,44(1):35-39 Wang G F, Li Q M, Qin X D, et al. Application of support-vector-machine in tool wear of multi-stage monitoring[J]. Journal of Tianjin University,2011,44(1):35-39 (in Chinese) [8] Suykens J A K, Vandewalle J. Least squares support vector machine classifiers[J]. Neural Processing Letters,1999,9(3):293-300 [9] Kalman R E. A new approach to linear filtering and prediction problems[J]. Journal of Basic Engineering,1960,82(1):35-45
点击查看大图
计量
- 文章访问数: 191
- HTML全文浏览量: 14
- PDF下载量: 11
- 被引次数: 0