基于神经网络的复合控制算法研究——哈尔滨理工大学学士学位论文.doc

基于神经网络的复合控制算法研究 摘要 本文从实际出发，以加热炉为研究对象，对受随机因素干扰的、具有大惯性、纯滞后的非线性分布参量的随机过程进行了研究。文中着重研究了神经网络模糊和PID控制的融合方式，了一种复合式控制方案——神经网络模糊PID控制算法，并将此算法与PID控制进行了仿真比较。 加热炉的温度控制属于典型的过程控制，因此我们首先想到的控制方案就是采用传统的比例积分微分（PID）调节器进行控制。根据被控对象的不同，适当地调整PID参数，可以获得比较满意的控制效果。然而，由于PID算法只是在系统模型参数为非时变的情况下，才能获得理想的效果。当一个调整好参数的PID控制器被应用到模型参数时变系统时，系统性能会变差。因此这种控制作用无法从根本上解决动态品质和稳态精度的矛盾。 本文所介绍的基于神经网络的复合控制算法，具有实时性强、抗干扰性好、控制精度高的优点，可应用于实际工业过程。在模糊控制中，模糊推理相当于对一种输入输出关系的映射，输入为前提，输出经非模糊化后即为推理的结果输出。利用神经网络的任意函数映射功能，就可以实现模糊推理，实现PID参数在线调整。综上可见，神经网络模糊PID控制必将有广阔的应用前景。 关键词　神经网络；模糊控制；PID控制；MATLAB The Research of the Composite Control Arithmetic Based on Neutral Network Abstract In view of practical use, taking the stove as object, the thesis studies the casual process of nonlinear and distributed parameter with big inertia, pure lag and casual disturbance. The feasibility and the form of combining neural network and fuzzy theory with PID control are discussed. The theory of compound control-fuzzy-PID control based on neutral network is put forward. In addition, comparison among the compound control and the PID control is introduced. The result indicates that this control arithmetic is superior to others in aspect of overshoot, steady-state error, anti-jamming etc. The temperature control of the stove is type of process control. So the control method first coming into our mind is the conventional PID control using PID adjustor. Then satisfactory effect will come out, through adjusting the PID parameter properly for the specific object. However, we can get the ideal effect only on the condition that the parameter of the system model is time-invariant .when a parameter is used to a time-variant system, the performance will become bad. So this control can not solve the conflict between dynamic quality and static precision. The compound control arithmetic introduced in this text is based on neutral network, and has the feature of real-time, good anti-jamming, high precision. So it can be used in practical industry process. In the c