基于模糊神经网络梁结构主动振动控制算法研究

发布时间：2018-09-01 16:00

【摘要】：压电材料作为一种智能材料，同时具备传感器与作动器的功能，由于其能够减轻结构质量，所以广泛应用于结构振动控制中。本文使用模糊控制，模糊神经网络（FNN）控制以及基于可变学习效率的FNN控制作为控制算法，针对粘贴有压电材料的悬臂梁进行主动振动控制研究。首先本文利用欧拉伯努利梁理论推导了压电悬臂梁的运动微分方程，对梁在自由状态下受到冲击激励时的时间位移曲线进行了仿真。然后，，阐述了模糊控制的基本理论和模糊控制器的设计方法由此设计了模糊控制器。由于模糊控制器不需要被控对象的精确数学模型，所以广泛应用于各种工业控制器中。但是模糊控制器不具备自学习能力，不能自动适应外界环境的改变，所以本文提出了模糊神经网络控制算法。模糊神经网络结合了模糊系统不需要被控对象的精确数学模型和神经网络具备自学的优点，在智能控制中受到技术人员的关注。然而，模糊神经网络收敛速度常常不理想，导致训练时间过长，不适用于实时自动控制系统。基于此问题，本文提出了一种基于可变学习速率的模糊神经网络优化方法。优化后的FNN控制系统在每次训练中都选择最优的学习速率，所以能够在系统允许最大误差前提下达到提高神经网络的收敛速度的效果。本文采用MATLAB来测试模糊系统，FNN系统，和优化后的FNN控制系统对压电悬臂梁的振动控制效果，通过仿真，可以得到各算法在振动抑制中的控制效果。同时仿真说明基于可变学习速率优化方法能够加快网络收敛从而提高控制算法性能。
[Abstract]:As a kind of intelligent material, piezoelectric material has the functions of sensors and actuators. Because of its ability to reduce the structure quality, piezoelectric materials are widely used in structural vibration control. In this paper, fuzzy control, fuzzy neural network (FNN) control and FNN control based on variable learning efficiency are used as control algorithms to study active vibration control of cantilever beams with piezoelectric materials. In this paper, the differential equation of motion of piezoelectric cantilever beam is derived by using Eulerian Bernoulli beam theory, and the time displacement curve of piezoelectric cantilever beam subjected to shock excitation is simulated. Then, the basic theory of fuzzy control and the design method of fuzzy controller are expounded. Fuzzy controller is widely used in various industrial controllers because it does not need the precise mathematical model of the controlled object. However, the fuzzy controller does not have the ability of self-learning and can not adapt to the change of the external environment automatically. Therefore, a fuzzy neural network control algorithm is proposed in this paper. The fuzzy neural network (FNN) combines the advantages of the precise mathematical model of the fuzzy system without the object under control and the neural network with the advantages of self-study, so it has attracted the attention of technicians in intelligent control. However, the convergence speed of fuzzy neural networks is often not ideal, which leads to long training time and is not suitable for real-time automatic control systems. Based on this problem, a fuzzy neural network optimization method based on variable learning rate is proposed. The optimized FNN control system selects the optimal learning rate in each training, so it can improve the convergence speed of neural network under the premise of the maximum error allowed by the system. In this paper, MATLAB is used to test the vibration control effect of the fuzzy system and the optimized FNN control system for piezoelectric cantilever beam. Through simulation, the control effect of each algorithm in vibration suppression can be obtained. At the same time, the simulation shows that the algorithm based on variable learning rate optimization can accelerate the convergence of the network and improve the performance of the control algorithm.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TB535

【参考文献】