计量学报2025，Vol.46Issue(10)：1433-1445,13.DOI:10.3969/j.issn.1000-1158.2025.10.05
基于CoAtNet-LSTM模型的多传感器信息融合刀具磨损预测
Based on CoAtNet-LSTM Model Multi-Sensor Information Fusion for Proposed Tool Wear
摘要
Abstract
Proposed a tool wear prediction model based is proposed on the long short-term memory(LSTM)and CoAtNet.Features are extracted from the sensor signals in the time domain,frequency domain,and time-frequency domain,followed by outlier detection and processing of signal features using the Isolation Forest algorithm.These processed features are then fed into the prediction model to obtain tool wear predictions,with model hyperparameters optimized through the Hyperband algorithm.The proposed model is validated using the PHM2010 CNC milling tool dataset to verify prediction accuracy.Experimental results demonstrate that the coefficient of determination of the proposed model shows average improvements of 12.73%and 16.44%compared to the original CoAtNet and LSTM network models,respectively.关键词
几何量计量/刀具磨损/CoAtNet-LSTM模型/长短期时间记忆网络/Hyperband算法/孤立森林算法Key words
geometric measurement/tool wear/CoAtNet-LSTM model/long and short term temporal memory networks/hyperband algorithm/isolation forest引用本文复制引用
李亚,尚轩丞,王海瑞,朱贵富..基于CoAtNet-LSTM模型的多传感器信息融合刀具磨损预测[J].计量学报,2025,46(10):1433-1445,13.基金项目
国家自然科学基金(61863016) （61863016）
