<p>During the operation of the knitting underwear machine, unstable yarn tension can have adverse effects on the quality of the underwear fabric, such as uneven density and reduced resilience, which significantly diminishes the comfort and durability of the textile products. In response to the issues of slow tension response speed and low tension control accuracy in the traditional yarn feeding method of current knitting underwear machines, a tension control system has been designed for the yarn feeding process. The system employs an adaptive ADRC algorithm, which is based on fuzzy sliding mode control, to detect the real-time trend of yarn tension changes, adjust the real-time target speed of the yarn feeding motor, and then utilizes permanent magnet synchronous motor vector control technology to drive the motor operation. This reduces tension disturbances during the yarn feeding process, enhances the response speed during sudden tension changes, and achieves precise control of yarn tension. Finally, a tension controllable yarn feeding experimental platform was constructed to simulate the knitting underwear machine, and tests were conducted to evaluate the control effectiveness of the yarn feeding tension control system under various algorithm controls. The results show that the control system based on the adaptive ADRC algorithm reduces the amplitude of tension fluctuation in the yarn target tension stabilization phase by more than 33% compared to traditional PID and ADRC algorithms. The average response time is shortened by over 35%, respectively. These comparative data demonstrate that the system can enhance the stability of yarn tension control during the yarn feeding process while meeting the varying elasticity demands of different fabrics and the specific yarn feeding speed requirements of various knitting processes, ultimately ensuring fabric quality.</p>