Abstract:

Objective With the continuous advancement of intelligent manufacturing process in knitting department, automatic bobbin changing and thread continuation of yarn frame are technical difficulties to be solved urgently. Aiming at the problem that it is impossible to detect the capture state of head yarn and tail yarn when using mechanical knotter to complete yarn joint on yarn frame, which leads to the inability to realize intelligent manufacturing in the knitting workshop, this paper proposes a detection method based on image pixel point measurement.

Method Based on the working principle of mechanical knotting machine, a yarn detection and recognition mechanism which integrates the installation box, camera module and light source was proposed. The mechanism was fixed on the transparent tube between the fan and the suction nozzle. Through the small embedded module, images were collected and processed including pixel counting and signals output in real time. The developed technique facilitated the low-cost yarn capture state detection on the yarn frame.

Results After the knotter moves to the position near the end line, the image without yarn is collected, and the initial number of white pixels is obtained after processing. 200 Groups of data are randomly selected to obtain the curve (Fig.9). The abscissa is the number of tests, and the ordinate is the number of pixels. The initial value of the number of white pixels varies between 24 592 and 24 651, and the maximum variation is only 59. After getting the initial value of the number of pixels, the system controls to clear it to get the corresponding number of pixels when there is no yarn. Theoretically, the number of pixels after clearing is 0. However, due to a small amount of light leakage in the installation box and the high sensitivity of pixel measurement, there is still a numerical fluctuation. Therefore, 200 groups of data are randomly selected to obtain the curve (Fig.10). The number of pixels without yarn after zeroing varies from 0 to 55. Then the head line and tail line absorption experiments were carried out at 5 different positions. After the head line is captured, the number of pixels changes significantly, and is far greater than its maximum fluctuation value of 55. At the same time, the curve fluctuation amplitude is close to that in Figs.9 and 10, which proves that the head line capture state can be recognized by measuring image pixels. In addition, the position has a great impact on the number of pixels, with a range of 204-512. After the two yarns are captured successfully, the change trend of the number of pixels obtained is basically consistent with that of a single yarn, and the number of pixels corresponding to positions 2 to 5 changes significantly more than that of a single yarn, so the capture status of the head thread and tail thread can be detected and recognized normally. The number of pixel points corresponding to position 1 is less than or close to positions 3 to 5 (Fig.11), but there is still a significant difference compared with the number of pixel points of single yarn position 1 and two yarn position 1.

Conclusion In this paper, taking the automatic bobbin change and thread continuation of the circular weft frame as an example, a yarn absorption detection algorithm based on image pixels is proposed according to the working principle of the mechanical knotting machine, and a special installation box and embedded module for yarn detection are designed, which achieves the online real-time recognition function of the bobbin head yarn and tail yarn absorption status before knotting. At the same time, through the experimental test and demonstration application of single and multiple absorption of common yarns with different colors, it is verified that this method has the advantages of high detection sensitivity, small size, low cost, etc. In addition, the detection mechanism and method are also applicable to wire break detection and other fields, so it has good promotion and application value. However, in-depth research on yarn contour, broken thread detection, yarn specification identification and the impact of vibration on the identification effect will be necessary for future work to further improve its applicability and stability.

Key words: mechanical knotter, yarn detection system, image pixel measurement, camera module, embedded module, circular knitting machine