摘要PID(比例-积分-微分)控制器是最早实现实用化的控制器，距今已有至少 50 年的历史，目前仍然是工业中应用最广泛、最普遍的控制器。传统的 PID 控制器 简单易懂，使用过程中无需精确的物理系统模型，所以它成为最受欢迎且应用最 为普遍的控制器。然而常规的 PID 控制器往往存在参数整定欠佳、性能不良以及 对运行环境的适应性很差等缺点。另一方面，由于神经网络具有很强的非线性映 射能力、自学习的能力、联想记忆的能力、并且具有并行处理批量信息的能力以 及良好的容错性能,目前已有很多研究工作试图将神经网络与 PID 控制结合起来, 以进一步提高自动控制系统的可靠性和鲁棒性。81549
本文研究了基于神经网络自学习的 PID 控制算法，设计了基于神经网络的 PID 控制系统。该系统采用神经网络对系统的即时误差进行处理,并根据处理结 果修改 PID 控制系统的参数，实现提高控制效果的目标。
本文使用 MATLAB 软件对该系统在不同受控对象上的控制效果进行了仿真。 仿真结果表明，基于神经网络的 PID 控制系统能够在非线性或比较复杂的受控对 象上获得较为理想的控制效果。
毕业论文关键词 PID 控制 神经网络 PID MATLAB 仿真
ABSTRACT PID (Proportional - Integral - Derivative) controller is the first practical use of the controller。 It has a history of at least 50 years and is still the industry's most extensive and most universal controller。 Traditional PID controller is simple to understand and use。 It can be used without the accurate physics model of the process that needs to be controlled, so it has become the most popular and common controller。 However, conventional controller also has a few disadvantages including poor parameter tuning, poor performance, poor adaptability to the operating environment and a few other shortcomings。 On the other hand, since neural networks have strong abilities in a few different aspects, including nonlinear mapping, self-learning, associative memory, parallel processing of batch information and good fault tolerance, a lot of work tries to combine neural networks with PID control to further improve the reliability and robustness of automatic control systems。
This studies PID control algorithm based on self-learning neural network。 A PID control system based on neural networks is designed。 The system uses a neural network system for real time processing of errors and modifies the parameters of PID control system based on the results, to achieve the goal to improve the control effects。
This work uses MATLAB software to simulate the effects of the control system on a few different objects that need to be controlled。 The simulation results show that the neural network based PID control system is able to achieve ideal control effects on non-linear or complex controlled objects。
Keywords : PID control neural network based PID MATLAB simulation
目 录
第一章 绪论 1
1。1 课题研究背景 1
1。2 课题的研究意义 2
1。2。1 感知模式识别 2
1。2。2 具有容错和容差能力 2
1。2。3 神经网络在工作时具有高速度以及潜在的超高速 2
1。3 课题目前研究现状 3
1。3。1 采用神经元网络确定 PID 参数 MATLAB神经网络自学习的PID控制算法研究:http://www.youerw.com/zidonghua/lunwen_95352.html