Based on Adaptive Dynamic Programming Carbon Fiber Diagonal Loom Tension Control Method

In order to solve the matters of poor tension control performance and weak anti-interference in warp tension control of carbon fiber diagonal loom, a tension control strategy based on adaptive dynamic programming is proposed. Above all, the let-off system tension control model is established after analyzing the weaving principle of carbon fiber loom. It considers the influence of opening mechanism. In the second place, based on the framework of ADHDP, BP neural network is utilized to design the action network, and wavelet neural network is used to design the critic network to enhance the ability of ADHDP to approach approximate and optimal control. Finally, the simulation control experiment of the tension system is carried out by using Matlab software. Compared with the classical ADHDP tension control, The action network and critic network are all designed by BP neural network. In the anti-interference response of the tension system state, the proposed tension control strategy shortens the response time by nearly 30.76 % and the overshoot by almost 23.3%. In the tracking response of the tension system state, the response time is shortened by nearly 32.9% and there is almost no overshoot. WNN-ADHDP tension control method improves the control accuracy and anti-interference ability of the tension system.