Akademik Veri Yönetim Sistemi
CIRP JOURNAL OF MANUFACTURING SCIENCE AND TECHNOLOGY, cilt.64, ss.160-170, 2026 (SCI-Expanded, Scopus)
This paper presents an Extended Kalman Filter (EKF)-based real-time methodology for in-process modal identification and tool–workpiece engagement detection in milling operations. Unlike conventional offline Experimental Modal Analysis (EMA) and Operational Modal Analysis (OMA) methods, the proposed approach enables online tracking of time-varying modal parameters without requiring external excitation and force measurement. It is proposed to utilize the acceleration measurements occurring between two consecutive tooth-workpiece engagements, for that purpose an EKF, based on the free vibration response model is constructed. The angular tool position is not measured, to identify the onset of free vibration, a robust engagement detection algorithm is developed, which remains effective under spindle speed variations and geometric inaccuracies. The complete framework consists of three stages: (i) recursive estimation of dominant modal parameters — natural frequency, damping ratio, and amplitude — using the EKF; (ii) adaptive engagement detection through thresholding of the mean absolute scaled error (MASE); and (iii) refinement of the estimated modal parameters via median and Kalman filtering to suppress Bernoulli and Gaussian noise. The proposed method is experimentally validated on a thin wall cantilever workpiece during end-milling where results are compared with conventional hammer-test EMA results. The identified modal parameters closely match the EMA results demonstrating the method’s potential in monitoring machining processes.