关键词: 纱线张力检测, 自适应, 张力波动, 信号优化, 干扰信号处理

Abstract:

Objective The objective of this research is to develop an adaptive Loess based data processing algorithm for mitigating noise interference in dynamic yarn tension signals. The study aims to address three types of noise, which are Singularity signal, low frequency coupling signals below 2.6 kHz, and high frequency interference signals above 2.6 kHz. By combining limit filtering, low pass filtering, and Loess smoothing techniques, the proposed algorithm seeks to achieve stability and accurate smoothing of yarn tension signals. The importance of this work lies in enhancing yarn tension stability, improving production efficiency, and preventing yarn defects, so as to provide a feasible solution for optimizing text systems and processes.

Method The research method employs an adaptive Loess data processing algorithm to address the noise interference in dynamic yarn tension signals. Three types of noise (abnormal hidden changes, low frequency coupling, and high frequency interference) are identified based on noise type, tension signal characteristics, and filtering methods. The proposed algorithm combined limit filtering, low pass filtering, and loess smoothing to achieve noise suppression and signal smoothing. An experimental platform was set up to validate the yarn tension measurement method, by comparing a yarn package, yarn feeder, tension sensor, hooks, and groove cylinder. Crochet hook mimics the work of a high-speed seamless lingerie machine with sensors for experimental data collection on yarns.

Results The method proposed in this article has achieved significant results in yarn tension measurement. The data was analyzed and the effectiveness of three different algorithms (SG, Adj, and Loess) was evaluated. The Loess algorithm was found most effective in achieving the smoothing of tension signals and the impact of response time. The effect of window width on tension signals was investigated experimentally, with a focus on the tension response time during signal processing. Window width appeared to be an important parameter in smoothing algorithms because it affects the trade-off between signal smoothness and response time. The response time of the loess algorithm increased with the increase of window width, and it was demonstrated that a wider window would lead to smoother signal, but with slower response to tension changes. By analyzing the inflection points, the optimal window width of the Loess algorithm was determined to be 120, where the signals achieved the highest smoothness while maintaining an acceptable response time.

In order to further evaluate the effectiveness of the Loess algorithm, the processed tension signal was compared with the original tension waveform. The Loess algorithm successfully filtered out noise interference while accurately representing the original tension waveform. The signal processing results were compared with the original data, and the signal to noise ratio (SNR) was calculated to evaluate the filtering effect. The adaptive Loess algorithm was proven to effectively smooth tension fluctuations, and in all three cases under consideration, the signal-to-noise ratio of yarn tension signals was improved by 25.014%, 27.661%, and 25.276%, respectively. The results showed that the Loess algorithm achieves the highest signal-to-noise ratio for all three types of tension signals, and it effectively reduces noise interference while maintaining the characteristics of the signal, providing a smoother tension waveform. Overall, the research results confirmed the practical feasibility of the proposed adaptive Loess weighted regression yarn tension optimization method. The Loess algorithm is believed to be the best choice for smoothing tension signals due to its excellent noise reduction performance and minimal impact on response time.

Conclusion Through this study, we have successfully explored the impact of yarn tension variation on textile quality and proposed a yarn tension signal optimization method based on Loess weighted regression. Experimental results demonstrate that the yarn tension signal after Loess processing outperforms other algorithms, achieving a higher signal-to-noise ratio (32.186 dB). This validates the feasibility of the method for real-time yarn tension measurement and control in practical working environments. This research holds significant practical implications for the textile industry. Accurate real-time yarn tension measurement contributes to improving textile quality stability and production efficiency. Moreover, by optimizing yarn tension signals, the negative impact of yarn tension on knitted fabric quality can be minimized, reducing the possibility of producing defective products. Future research can further apply this method to industrial knitting machines and integrate it with other advanced technologies to enhance the accuracy and stability of yarn tension measurement. Additionally, exploring deeper relationships between yarn tension signals and textile quality will help optimize production processes and elevate textile quality. In summary, this study provides an effective solution for yarn tension measurement, fostering quality control and technological advancements in the textile sector. Continuous improvement and application of this method will lead the textile industry towards higher quality and greater efficiency.

Key words: yarn tension detection, odaptive, tension fluctuation, signal optimization, interference signal processing