Abstract:

Objective During textile production, yarn tension has an important impact on product quality and production efficiency, and it is necessary to monitor the peak value of yarn tension in real time and obtain the stable value of current yarn tension. The textile equipment measures the yarn tension transmitted by the yarn conveyor in real time through the yarn tension sensor, and transmits the measurement results to the yarn conveyor for regulation. The existing yarn tension sensor can monitor the yarn tension, but because the measured tension signal has problems such as non convergence, large chattering, low accuracy and many burrs, it cannot be applied to the feedback regulation of yarn tension. Otherwise, the measurement error and amplitude will become larger.

Method When the yarn is fed and knitted into a loop by the yarn feeder and the needle cylinder at a certain speed ratio, the tension will fluctuate with high frequency while keeping relatively constant under the dynamic traction and transportation of the knitting needle and the yarn feeder. On the one hand, the yarn drawn by the knitting needle is a process of intermittently variable speed yarn bending and looping, with periodic motion characteristics. On the other hand, the yarn itself has certain elasticity. Under the action of periodic traction, the stress wave in the yarn body will also be scattered and reflected during the transmission process. The dominant performance is the high-frequency fluctuation of yarn tension. Taking the high-speed seamless underwear machine as an example, the highest frequency of yarn tension measurement is 2.6 kHz, so the tension sensor should meet the highest frequency response requirements. At the same time, on the premise that the yarn tension measurement does not distort and amplify the effective signal, the signal higher than this frequency should be filtered out. This paper first analyzes the characteristics of the problems existing in the existing yarn tension sensors, divides the interference signals into three types, namely, low frequency, high frequency and singular point noise, and then designs three algorithms to deal with them: first, filter the singular noise through the amplitude limiting filtering algorithm (the amplitude of the tension measurement results at a moment is far greater than the actual tension measurement results). High frequency noise is removed by low-pass filtering, and the yarn tension working frequency is taken as the threshold value. The signal above this frequency is defined as high frequency signal; The S-G convolution algorithm is used to remove the low-frequency coupling noise. The low-frequency even coupling signal and the actual working frequency are intertwined, and the low-frequency signal cannot be removed separately. In order to verify the reliability and practicability of the algorithm, an experimental platform for yarn tension measurement was built to compare and analyze the yarn tension measurement data and standard data under constant tension and variable tension conditions.

Results In order to further verify the accuracy and effectiveness of the algorithm, the yarn conveyor of the control platform changed the yarn tension from 36 cN to 42 cN, and used TENSOMETRIC tension sensor and the sensor designed in this paper to conduct real-time testing of tension fluctuation. It can be seen that the actual measured yarn tension conforms to the mutation law, and the yarn mutation tension fitted by this method is better than the tension fitted by the standard sensor. The results show that this method can control the tension measurement error within 0.6%, which has guiding significance for the yarn tension detection and control system under complex working conditions.

Conclusion By using the cantilever structure, filtering and convolution algorithm, this paper proposes a contact yarn tension detection method based on convolution filtering algorithm. Firstly, according to the characteristics of the yarn fluctuation of the high-speed seamless underwear machine and the structural characteristics of the cantilever beam, the optimization method of the yarn tension sensor is determined. Then, the acquired data are processed by amplitude limiting filtering, low-pass filtering and S-G convolution algorithm. Finally, the measured results are compared with those of TENSOMETRIC tension sensor through experiments. The tension error is within 0.6% and the standard deviation is within 0.62%. Moreover, this method has good applicability to the detection of yarn sudden change tension, and can meet the real-time measurement of yarn tension under complex working environment.

Key words: convolution filtering, cantilever beam, piezoresistive effect, yarn tension, noise interference, real-time measurement