基于自适应积分补偿神经网络模糊气压解耦控制

Fuzzy Air Pressure Decoupling Control Based on Adaptive Integral Compensation Neural Network

摘要: 多支路气压系统中的各支路气压的变化不仅会影响其他支路气压的变化，还会影响总管气压的变化，导致各支路气箱中的气压与总管的气压产生了耦合震荡。针对气压控制系统中多支路气压的耦合震荡的问题，提出了自适应积分补偿神经网络模糊控制方法对气压控制系统进行解耦控制，采用自适应神经网络对模糊规则进行训练和学习，并引入了自整定积分控制器共同控制多路高速开关电磁阀，达到多路气压精准跟踪控制效果。试验平台采用AMESim搭建了64支路的气压控制系统仿真模型，并结合MATLAB/Simulink进行联合仿真分析。仿真实验结果表明，所提出的自适应积分补偿神经网络模糊控制算法对气压稳定控制效果提升明显，在气压控制系统中具有控制精度高、鲁棒性好、稳定性强等特点。

Abstract: The change of air pressure in each branch of the multi branch air pressure system will not only af-fect the change of air pressure in other branches, but also affect the change of air pressure in the main pipe, resulting in a coupling vibration between the air pressure in each branch air box and the air pressure in the main pipe. Aiming at the problem of coupling vibration of multi branch air pressure in the air pressure control system, this paper proposes an adaptive integral compensation neural network fuzzy control method to decouple the air pressure control system, uses the adaptive neural network to train and learn the fuzzy rules, and introduces a self-tuning integral controller to jointly control the multi-channel high-speed switch solenoid valve, so as to achieve the effect of mul-ti-channel air pressure accurate tracking control. The test platform uses AMESim to build the simu-lation model of 64 branch air pressure control system, and combines MATLAB/ Simulink for joint simulation analysis. The simulation results show that the proposed adaptive integral compensation neural network fuzzy control algorithm can significantly improve the air pressure stability control effect, and has the characteristics of high control accuracy, good robustness and strong stability in the air pressure control system.