Research On Network Control Strategy Of Carbon Fiber Angle Loom

With the rapid development of economy and intelligent technology,the carbon fiber materials have been developed in civil manufacturing,aerospace,mechanical equipment and military industry due to the characters of light weight,high modulus,high temperature resistance and fatigue resistance.Therefore,the carbon fiber equipment system with high efficiency is becoming more and more popular among researchers.In order to improve the quality of fabric in complex nonlinear networked control system and eliminate the influence of external interference in the process of weaving,the carbon fiber diagonal loom is regarded as the research object and the control strategy of the system is deeply studied around the yarn tension fluctuation.The principle of carbon fiber diagonal loom is analyzed and the influence of each subsystem of the loom is studied by this paper at first.Then the let-off system and the opening system are combined to build a tension control model.The influence of various interferences such as communication protocol,network delay,data loss,sampling period and node driven is studied deeply in order to offer a theoretical foundation for the project of the follow-up controller.In networked control system,a single neuron adaptive controller is designed by this paper in view of the poor robustness and weak anti-interference ability of PID controller.The performance index function combined with the idea of dynamic programming is introduced to minimize the system error under the given constraint conditions and the Lyapunov stability proof is given.Combined with the mathematical model of tension,a simulation experiment platform based on True Time and MATLAB is built.The results show that only considering the network induced delay,the traditional PID control strategy can tolerate up to 45 ms,only considering the data loss,the maximum can tolerate 20% data loss.The maximum allowable network interference of the improved single neuron adaptive control strategy is 70 ms delay and 40% data loss respectively,and it has smaller overshoot when the two kinds of interference exist at the same time.BP neural network control strategy is designed for the situation that the overshoot of the improved single neuron adaptive control strategy is too large under the condition of increasing network induced delay and data loss rate.In view of BP neural network is easy to stop training when it is caught in the local minima,the inertia coefficient is introduced in the error adjustment process to achieve the aim of jumping out of the dead cycle.The results show that BP neural network has a little delay in response speed,but it can tolerate 0.6s delay and 55% data loss rate.Compared with the improved single neuron adaptive control strategy,BP neural network has more gentle tension fluctuation under the condition of greater network interference.It is more suitable for the yarn tension control of carbon fiber diagonal loom under the condition of large interference in actual production.