基于RBF神经网络整定的PID控制器设计及仿真_毕业设计(论文).docVIP

基于RBF神经网络整定的PID控制器设计及仿真 摘 要 目前，因为PID控制具有简单的控制结构，可通过调节比例积分和微分取得基本满意的控制性能，在实际应用中又较易于整定，所以广泛应用于过程控制和运动控制中，尤其在可建立精确模型的确定性控制系统中应用比较多。然而随着现代工业过程的日益复杂，对控制要求的逐步增高（如稳定性、准确性、快速性等），经典控制理论面临着严重的挑战。对工业控制领域中非线性系统，采用传统PID 控制不能获得满意的控制效果。采用基于梯度下降算法优化RBF神经网络，它将神经网络和PID控制技术融为一体，既具有常规PID控制器结构简单、物理意义明确的优点，同时又具有神经网络自学习、自适应的功能。因此，本文通过对RBF神经网络的结构和计算方法的学习，设计一个基于RBF神经网络整定的PID控制器，构建其模型，进而编写M语言程序。运用MATLAB软件对所设计的RBF神经网络整定的PID控制算法进行仿真研究。然后再进一步通过仿真实验数据，研究本控制系统的稳定性，鲁棒性，抗干扰能力等。 关键词：PID；RBF神经网络；参数整定 SETTING OF THE PID CONTROLLER BASED ON RBF NEURAL NETWORK DESIGN AND SIMULATION Abstract At present, because the PID control has a simple control structure, through adjusting the proportional integral and differential gain basic satisfactory control performance, and is relatively easy to setting in practical application, so widely used in process control and motion control, especially in the accurate model can be built more deterministic control system application. With the increasingly complex of the modern industrial process, however, increased step by step to control requirements (e.g., stability, accuracy and quickness, etc.), classical control theory is faced with severe challenges. Non-linear systems in industrial control field, using the traditional PID control can not obtain satisfactory control effect. Optimized RBF neural network based on gradient descent algorithm, it will be integrated neural network and PID control technology, with a conventional PID controller has simple structure, physical meaning is clear advantages, at the same time with neural network self-learning, adaptive function. Therefore, this article through to the RBF neural network structure and the calculation method of learning, to design a setting of the PID controller based on RBF neural network, constructs its model, and then write M language program. Using the MATLAB software to design the RBF neural network setting of PID control algorithm simulation research. Data and then further through simulation experiment,