Noncontact tension detection method for carbon fiber precursor filaments based on laser Doppler vibrometer and optimized variational mode decomposition

Tension detection during the spinning process of carbon fiber precursor filaments plays a crucial role in enhancing the quality of carbon fibers. The noncontact tension detection method is proposed for carbon fiber precursor filaments in this paper. Based on Hamilton’s principle, the dynamic differential equations of the ribbonlike filament bundle are established, and a general formula relating filament bundle tension to vibration frequency is derived. In terms of signal processing methods, the Improved Aquila Optimization algorithm is first used to optimize the variational mode decomposition parameters, including the number of modes K and the penalty factor α, to decompose the signal into components of different frequencies. Subsequently, a combined screening method using the Pearson correlation coefficient, amplitude-sensitive arrangement entropy, and DBSCAN is employed to identify and remove noise signals, thereby achieving signal denoising. Experimental results indicate that this method for screening signal components achieves better results compared with the single correlation coefficient screening method. The measurement errors of laser-Doppler-vibrometer-based tension detection are within ±11.5%. This method enables noncontact and nondestructive tension detection, providing a new solution for situations where traditional contact-based detection is not feasible. Therefore, this method has significant research potential.