【作者】 朱庆华；

【导师】 林小峰；

【作者基本信息】 广西大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 神经网络具有并行处理能力、自学习、自适应能力和可以逼近任意的非线性函数等特点，是解决非线性、大滞后、多变量、不确定的复杂控制问题的一条有效的途径。它在控制领域中辨识、参数优化、控制等方面得到广泛应用。 本文就神经网络在辨识、参数优化和控制方面的应用展开研究。首先给出神经网络控制系统的整体设计思路，编写相应的软件，该软件包括多种神经网络网型，并且可以通过不同方式与组态软件集成运行。然后利用该软件对被控对象进行辨识、参数优化和控制：使用多种网型对三区加热炉进行辨识，通过比较辨识结果，得出采用Elman网络辨识动态系统效果较好的结论；接下来利用神经网络对PID控制器进行参数优化，实现基于神经网络的PID自适应控制，并对此控制系统改进，形成了有模型的神经网络PID自适应控制，获得较好的控制效果；最后利用PID神经网络直接作为控制器对三容水槽和立式电炉进行控制，结果表明，该方法有较好的控制效果。 上述研究表明神经网络能在一定程度上解决控制中的复杂问题，有较好的应用前景。更多还原

【Abstract】 The neural network has many abilities: parallel processing, self-learning and self-adapting, approaching any nonlinear function etc. ; to deal with the problem of control in the industry process with wide lag?nonlinear?multivariable and uncertainty, it is an effective way to develop neural network. So neural network is used widely in these aspects: identification, control, parameter optimization.The research of this dissertation lies on three aspects: identification, control, parameter optimization. Firstly this dissertation introduces the design of neural network control system. According to the design, we program the neural network software. The software includes many kinds of networks, and can run with different configration software through different ways. With this software, we can construct the control system. Then we use some networks to identify the furnace. Comparingthe results, a conclusion can be drew: To identify the dynamic system , elman network can get better result. After that, BP neural network is used to optimize the parameters of PID controller, and the controller is also applied to the control of furnace temperature. Then we improved the system with model. At last, we use PID neural network to control the objects directly. The result of the control in three-tank liquid level equipment and stand-up furnace is excellent.The researches above indicate that the neural network can solve the complicated problem of control to a certain extent, and has better prospect in the application of control.更多还原