基于CAFSA-FNN的扭矩加载系统控制研究  

Torque loading system based on modified fuzzy neural network

作　　者：白国振[1] 朱灵康[1] 杨雷[1] 周媛[1] 

机构地区：[1]上海理工大学机械工程学院,上海200093

出　　处：《计算机应用研究》2017年第7期2072-2076,共5页Application Research of Computers

基　　金：上海市自然科学基金资助项目(13ZR1458500)

摘　　要：针对磁粉制动器扭矩加载系统的非线性和滞后性,提出了一种基于混沌人工鱼群—模糊神经网络(CAFSA-FNN)PID控制器。该控制器采用基于Mamdani模型的模糊神经网络来整定PID控制器的控制参数,并结合混沌人工鱼群算法离线粗调和BP算法在线细调来学习和调整模糊神经网络的参数。利用MATLAB进行离线仿真优化,在此基础上使用PID控制器、模糊神经网络控制器、人工鱼群—模糊神经网络控制器以及设计的CAFSA-FNN控制器进行磁粉制动器扭矩加载实验,实验结果证明了该控制器的稳定性、快速性和有效性,能够解决滞后性问题。Considering the nonlinear and uncertainty of magnetic powder brake（MPB） torque loading system, this paper pre-sented a new fuzzy neural network （FNN） PID controller based on chaotic artificial fish swarm algorithm（CAFSA）. The controller used a fuzzy neural network based on Mamdani model to adjust the control parameters of PID controllers.It also combined offline chaotic artificial fish swarm algorithm and online BP algorithm to learn and adjust the parameters of fuzzy neural network. Using MATLAB for offline simulation and optimization, then did experiments using the PID controller, FNN controller, AFSA-FNN controller and the designed controller. The experimental results show that this controller is more stability, faster and more effective.It can solve the lag problem of the MPB torque loading system.

关 键 词：磁粉制动器 扭矩加载 模糊神经网络 人工鱼群 混沌系统 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]