为更好地解决传统PID对负压供墨系统温度、负压值精度低、缺乏自适应性、跟随性能差等问题,本文将RBF神经网络与传统PID控制算法结合起来,实现动态辨识,通过利用神经网络的学习能力,可以根据控制环境在线修正PID控制的比例、积分、微分...为更好地解决传统PID对负压供墨系统温度、负压值精度低、缺乏自适应性、跟随性能差等问题,本文将RBF神经网络与传统PID控制算法结合起来,实现动态辨识,通过利用神经网络的学习能力,可以根据控制环境在线修正PID控制的比例、积分、微分参数,使其更加符合工业需求,从而能够提升系统的实时性以及适应性,通过加入阶跃信号,基于MATLAB软件中的Simulink环境对控制系统进行仿真。通过对比检验传统PID控制算法与模糊PID控制算法。经过仿真测试结果得出结论:基于RBF神经网络优化的PID控制算法具有响应速度快、超调小等优点,解决了控制温度与负压值过程中滞后和耦合大的问题,显著改善负压供墨系统性能。To better address the issues of traditional PID control in negative pressure ink supply systems, such as low temperature and pressure accuracy, lack of adaptability, and poor tracking performance, this paper integrates the RBF neural network with the traditional PID control algorithm to achieve dynamic identification. By leveraging the learning capability of the neural network, the proportional, integral, and derivative parameters of the PID control can be adjusted online according to the control environment, making it more suitable for industrial needs. This enhancement improves the system’s real-time response and adaptability. A step signal was introduced, and the control system was simulated in the Simulink environment of MATLAB software. By comparing the traditional PID control algorithm with the fuzzy PID control algorithm, simulation test results concluded that the PID control algorithm optimized by the RBF neural network offers advantages such as fast response speed and minimal overshoot, effectively solving the problems of lag and significant coupling in temperature and pressure control, thereby significantly improving the performance of the negative pressure ink supply system.展开更多
文摘为更好地解决传统PID对负压供墨系统温度、负压值精度低、缺乏自适应性、跟随性能差等问题,本文将RBF神经网络与传统PID控制算法结合起来,实现动态辨识,通过利用神经网络的学习能力,可以根据控制环境在线修正PID控制的比例、积分、微分参数,使其更加符合工业需求,从而能够提升系统的实时性以及适应性,通过加入阶跃信号,基于MATLAB软件中的Simulink环境对控制系统进行仿真。通过对比检验传统PID控制算法与模糊PID控制算法。经过仿真测试结果得出结论:基于RBF神经网络优化的PID控制算法具有响应速度快、超调小等优点,解决了控制温度与负压值过程中滞后和耦合大的问题,显著改善负压供墨系统性能。To better address the issues of traditional PID control in negative pressure ink supply systems, such as low temperature and pressure accuracy, lack of adaptability, and poor tracking performance, this paper integrates the RBF neural network with the traditional PID control algorithm to achieve dynamic identification. By leveraging the learning capability of the neural network, the proportional, integral, and derivative parameters of the PID control can be adjusted online according to the control environment, making it more suitable for industrial needs. This enhancement improves the system’s real-time response and adaptability. A step signal was introduced, and the control system was simulated in the Simulink environment of MATLAB software. By comparing the traditional PID control algorithm with the fuzzy PID control algorithm, simulation test results concluded that the PID control algorithm optimized by the RBF neural network offers advantages such as fast response speed and minimal overshoot, effectively solving the problems of lag and significant coupling in temperature and pressure control, thereby significantly improving the performance of the negative pressure ink supply system.
