基于改进型模糊神经网络PID阀位控制研究

Research on position control of valve based on improved fuzzy neural network PID

针对调节阀控制系统在实际生产中存在的大滞后、非线性等问题,提出一种改进粒子群算法优化的模糊神经网络比例积分微分(PID)控制模型用于阀位控制,该模型利用模糊神经网络的自学习能力,实现对PID控制参数的实时在线整定,并且通过将改进粒子群算法与BP算法相结合的方式,实现对模糊神经网络参数的粗调和细调,克服了模糊神经网络收敛缓慢、易陷入局部最优的缺点;最后,利用MATLAB和AMESim软件进行联合仿真,仿真结果表明,该模型相比于其他两种算法在调节时间、超调量等性能方面都有很大的提升,并且表现出更强的鲁棒性和抗扰动能力,能够使阀位控制更加稳定可靠。

Aiming at the problems of large lag and nonlinearity in the actual production of the control valve control system,a fuzzy neural network PID control model optimized by the improved particle swarm optimization algorithm was proposed for valve position control.This model uses the self-learning ability of the fuzzy neural network to realize real-time online tuning of PID parameters,and through the combination of improved particle swarm algorithm and BP algorithm,it realizes the coarse adjustment and fine adjustment of the fuzzy neural network parameters,which overcomes the shortcomings of slow convergence of fuzzy neural network and easy to fall into local optimum.Finally,MATLAB and AMESim software were used for co-simulation.The simulation result shows that compared with the other two algorithms,the model has a great improvement in performance such as adjustment time and overshoot,and it shows stronger robustness and anti-disturbance ability,which can make the valve position control more stable and reliable.