摘要
针对铣床轴承沟道磨损检测精度较低的问题,提出改进K均值聚类下铣床轴承沟道磨损检测方法。通过UT372手持式光电速度仪与MPU-605压电加速度仪,采集铣床轴承沟道磨损数据。根据最远最近原则,初步选取数据集的聚类中心。使用欧氏距离计算出数据集各点间的距离与所有数据的平均距离,并结合交叉验证确定聚类中心的两个阈值。引入Canopy算法改进K均值聚类,确定全局最佳的聚类中心,从而实现铣床轴承沟道磨损的智能检测。试验结果表明,改进K均值聚类算法在铣床轴承沟道磨损检测中,迭代次数固定为15次、Jaccard系数极其接近1。该方法能够显著提升聚类的计算速度和稳定性,可识别不同铣床轴承沟道磨损故障类型,且检测精度高。
Aiming at the problem of low accuracy of bearing groove wear detection of milling machine,an detection method of bearing groove wear of milling machine under improved K-mean clustering is proposed.Through the UT372 handheld photoelectric velocimeter and MPU-605 piezoelectric accelerometer,the milling machine bearing groove wear data are collected.The clustering center of the dataset is initially selected based on the farthest-nearest principle.The distance among the points in the data set and the average distance of all the data are calculated using the Euclidean distance,and the two thresholds of the clustering center are determined by combining with cross-validation.Canopy algorithm is introduced to improve the K-mean clustering,to determine the global best clustering center and realize the intelligent detection of the bearings groove wear of milling machine.The experimental results show that the improved K-mean clustering algorithm in the detection of milling machine bearing groove wear,the number of iterations is fixed to 15 times,Jaccard coefficient are extremely close to 1.This method can significantly improve the computational speed and stability of the clustering and can identify different types of milling machine bearing groove wear faults,and the detection accuracy is high.
作者
睢雪亮
夏景攀
SUI Xueliang;XIA Jingpan(Transportation College,Henan Technician Institute,Zhengzhou 450000,China)
出处
《自动化仪表》
CAS
2024年第10期80-85,共6页
Process Automation Instrumentation
基金
河南省终身教育课题和课程开发基金资助项目(豫教[2023]70037)。
关键词
铣床轴承
沟道磨损
K均值聚类
Canopy算法
聚类中心
欧氏距离
最远最近原则
交叉验证确定
Milling machine bearing
Groove wear
K-mean clustering
Canopy algorithm
Cluster center
Euclidean distance
Farthest-nearest principle
Cross-validation determination
