Traditional proportional–integral–derivative (PID) control performance optimization is an essential method to improve a loom’s warp tension control performance. This work proposes an improved genetic algorithm optimized PID control scheme to overcome the decline in control performance of the traditional PID control algorithm in a loom’s warp tension control system. Through the decoupling analysis of loom motion mechanism, the establishment of warp tension model and the optimization of fitness evaluation mechanism of genetic algorithm can effectively overcome the problems of local optimal solution and algorithm degradation of genetic algorithm. Simulation experiments were carried out with the traditional PID, the integral separation PID, and the genetic PID in warp tension control. The results show the advantage of the genetic-PID algorithm to control warp tension stability. Ultimately, according to the functional characteristics of the loom mechanism, a tension control platform for experimental studies was established. The test results show that the maximum fluctuation range of warp tension is within [−2, +6] at the test speed of 850 rpm, which meets the requirements of long-term stable and reliable control of warp tension under different weaving conditions.