Implementing Kalman Filtering Algorithms for Estimating Clamp Force on a Test Rig : Testing the Power and Limitations of Unscented Kalman Filter-based Estimations

Sammanfattning: his study explores clamp force estimation using Unscented Kalman Filtering (UKF) in torque-controlled tightening scenarios with various velocity profiles. Previous research has explored the impact of velocity levels on target torque and clamping force, but only using hand-held tools. Prior research is extended by implementing UKF in a fixed setup, using the QST42, to remove user errors. Four strategies, Continuous Drive, TurboTight, Accelerating Tightening, and Paused Tightening, are analyzed using error and quality factor metrics. In Continuous Drive, both hand-held and fixed rigshave mean errors of approximately 4.09% and 4.14%, with quality factors of 88.38% and 97.72%.UKF adapts well in TurboTight, with mean errors of 3.50% (hand-held) and 5.23% (fixed rigs), and quality factors of 93.02% and 94.44%, respectively. Dynamic strategies like Accelerating Tightening- yield higher mean errors (10.33%) and quality factors (94.86%), while Paused Tightening results in a mean error of 5.17% and a quality factor of 76.86%. Tailoring UKF calibration is crucial for accuracy. Overall, this research underscores the close correlation between UKF’s performance and the dynamics of the tightening strategy. The implications extend to industrial applications, advocating for strategy-specific adjustments to enhance clamp force estimation accuracy. This study contributes to advancing UKF’s applicability in real-world scenarios, providing a foundational framework to enhance the accuracy and reliability of clamp force estimations.