机械加工
公制(单位)
刀具磨损
估计员
概化理论
卡尔曼滤波器
计算机科学
性能指标
工程类
数据挖掘
人工智能
统计
机械工程
数学
运营管理
管理
经济
作者
Qian Yang,Debasish Mishra,Utsav Awasthi,George M. Bollas,Krishna R. Pattipati
标识
DOI:10.1016/j.jmsy.2024.04.001
摘要
This article presents a monitoring method developed for diagnosis and prognosis of cutting tools in precision machining processes. Diagnosis here pertains to estimating the flank wear of cutting tools, while prognosis forecasts their remaining useful life (RUL). In-house machining tests were conducted on three machines with varying machining parameters to develop a generalized monitoring method. The method operates in two steps. The first step involves signal processing to extract the health indicators from vibration signals. A distance metric is applied to statistical signal features, and the evolution of this metric over machining runs is computed. The distance metric is a strong indicator of progressive tool wear, with a Pearson correlation coefficient higher than 0.85, averaged across the three machines. The second step is modeling, where health indicators are used as measurements in an Interacting Multiple Model (IMM) estimation framework with Extended Kalman Filters (EKF) serving as component estimators for diagnosis and prognosis, namely assessing the tool wear and forecasting the RUL. The generalizability of the method is ensured by considering training and testing samples of machining data coming from separate machines studied in this work. The tool wear and RUL are predicted with a mean absolute error (MAE) of 0.0522 mm and 1.7091 runs, respectively. The application of the monitoring method is also shown on the IEEE PHM 2010 data for benchmarking. The results demonstrate the efficacy and practicality of the proposed method.
科研通智能强力驱动
Strongly Powered by AbleSci AI