A Study on the Irregularity of Sliver Structure Based on Computer Image Recognition Technology

The structural irregularity of slivers significantly impacts both the spinning process and yarn quality, which is usually characterized by the variance-length curve. Conventional methods for determining this curve, such as the cutting and weighing technique, are plagued by inefficiency during continuous slivers or yarns irregularity testing. In this paper, a novel method for determining the variance-length curve was proposed based on computer image recognition technology, demonstrating superior accuracy and efficiency compared to traditional methods. Utilizing this technique, we investigated key factors influencing the variance-length curves of slivers and rovings. Results indicate that sliver irregularity, spinning process parameters, and testing length collectively affect structural characteristics. It is noteworthy that in the analysis of roving irregularity, the variance-length curve tends to stabilize when the testing length exceeds 8 m. The poorer the roving quality, the longer the testing length needed. For sliver irregularity analysis, however, the variance-length curve stabilizes when the testing length surpasses 4 m. Furthermore, exponential fitting (base e) effectively modeled the discrete points of the variance-length curve, achieving high precision in capturing experimental data. These findings provide a theoretical foundation for fiber assembly deconstruction, spinning process optimization, and virtual spinning quality prediction.