Abstract:

Objective Warp sizing is a key process for improving weaving efficiency， in which the pressure distribution in the squeezing zone plays a decisive role in size penetration and membrane formation. However， the conventional methods， such as the indentation method， can only estimate squeezing width without being able to characterize the actual pressure distribution. To address this limitation， this study proposes a pressure detection method， based on the use of a thin-film pressure sensor， to characterize the pressure distribution in the squeezing zone.

Method A pressure data acquisition system was developed， consisting of a thin-film pressure sensor， a linear voltage conversion module， an Arduino Uno R3， and a host computer data acquisition software. After sensor selection and system calibration， the sensor was mounted on the surface of the squeezing roller to measure the pressure under different levels of squeezing force and hardness and thickness of the rubber covering. The acquired signals were processed using a mean filter for noise reduction. A one-dimensional Gaussian function was then employed to fit the pressure data， from which two characteristic indicators， namely maximum pressure and squeezing time， were extracted to represent the pressure distribution.

Results The proposed method captured the pressure distribution in the squeezing zone with success under various operating conditions. The CGQ-5 thin-film pressure sensor demonstrated better stability and lower noise than the CGQ-3 sensor， as evidenced by smaller standard deviations in both analog-to-digital converter （ADC） values and voltage signals. For example， in the no-load drift test， the standard deviations of ADC and voltage for the CGQ-5 sensor were 3.69 and 18.06， respectively， compared with 4.86 and 23.74 for the CGQ-3 sensor. Similar advantages were observed in the no-load noise and constant-pressure noise tests. After calibration， the effective voltage range of 110 - 3 300 mV corresponded to a pressure range of 0.294-98.07 N， ensuring reliable voltage-to-pressure conversion. Signal processing results showed that the mean filter achieved a standard deviation of 661.07， a Pearson correlation coefficient of 0.918， and a peak retention of 82.63%， indicating an optimal balance between noise reduction and dynamic response. The one-dimensional Gaussian function fitting achieved coefficients of determination （R²） higher than 0.95， demonstrating that the pressure distribution was well characterized. Compared with the indentation method， the proposed approach was capable of distinguishing subtle differences in pressure distribution. Further analysis revealed that both maximum pressure and squeezing time increased linearly with squeezing force， with average R² values of 0.961 9 and 0.983 9， respectively. Increasing rubber covering hardness led to higher maximum pressure but reduced squeezing time， whereas increasing rubber covering thickness resulted in a nonlinear trend， with maximum pressure decreasing first and then increasing， and squeezing time showing the opposite tendency. These results are consistent with theoretical expectations， further supporting the capability of the proposed method to characterize the pressure distribution in the squeezing zone.

Conclusion A method for pressure detection in the squeezing zone of a sizing machine was developed based on the use of a thin-film pressure sensor. By integrating sensor selection， system calibration， signal processing， and one-dimensional Gaussian function fitting， the method enables accurate characterization of pressure distribution. The extracted indicators， maximum pressure and squeezing time， effectively reflect variations under different process conditions. Compared with conventional methods， the proposed method provides more comprehen

Key words: sizing machine, pressure distribution measurement, thin-film pressure sensor, squeezing force, squeezing roller rubber covering parameter, squeezing zone