Advanced Filtering Strategies for Residual Error Mitigation in Vibration Sensors for Precision Manufacturing

Vibration monitoring in machine tools is essential for ensuring precision and quality in industrial manufacturing. However, vibration sensors, including both general-purpose and MEMS-based sensors integral to Industrial Internet of Things (IIoT) systems which offer compact and cost-effective solutions for continuous monitoring, are prone to residual errors that persist even after application of systematic compensation and calibration techniques. These errors can degrade the accuracy of predictive maintenance and quality control systems, which are vital in the era of Industry 4.0 and smart manufacturing. This research investigates various filtering techniques to minimize these residual errors, with a focus on their applicability to the non-stationary and complex vibration signals typical in machine tool environments. Several adaptive filtering methods, including Savitzky-Golay (SG), Wiener filtering, Wavelet denoising, Adaptive Recursive Least Squares (RLS), and Kalman Filtering (KF), were evaluated using a simulated dynamic noisy vibration signal representative of an industrial CNC machine. The evaluation criteria included Signal-to-Noise Ratio (SNR) improvement, Mean Squared Error (MSE), and convergence time, ensuring real-time suitability for practical industrial applications.
Extensive Monte Carlo simulations were conducted to compare the effectiveness of these techniques in reducing noise and improving signal estimation accuracy. Significant differences were observed in their ability to manage the non-linear and non-stationary characteristics of machine tool vibrations. Advanced Kalman filtering techniques, in particular, showed potential for processing non-linear systems in vibration signal processing. The findings contribute to the field of precision engineering by offering a comprehensive comparison of filtering techniques and proposing advanced methods for residual error compensation in vibration sensors. This work has important implications for enhancing measurement accuracy, machine tool performance, and quality control in industrial manufacturing, while also improving IIoT-based condition monitoring and more precise predictive maintenance strategies, and overall optimization of smart manufacturing processes as they are dependent on high quality sensing.