模糊神经网络控制器用于电站主汽压控制的研究

锅炉的燃烧过程是一个多参数、非线性、时变以及变量强耦合的过程，很难建立被控对象的准确数学模型。根据主汽压被控对象的动态特性，设计了一个模糊神经网络自适应控制系统，引用模糊高斯基函数神经网络结构，并采用基于变尺度优化学习算法的改进型学习算法，其学习信号由神经网络辨识器(NNI)提供。利用神经网络的非线性映射能力，能很好的解决被制对象的动态特性具有非线性、时变性、参数可变等问题。仿真对比试验表明，主汽压控制系统引入模糊神经网络控制器(FNNC)后，系统的响应速度变快，调节精度提高。该控制器的适应性、鲁棒性也明显优于常规PID控制器。

Application of fuzzy neural networks controller to the main steam pressure control in power station

The process of boiler's combustion is a multi-parameter, nonlinear, time-varying and strongly coupling course, so the exact mathematical models of the controlled objects are hardly constructed. According to dynamic characteristic of the main steam pressure, a fuzzy-neural-networks self-adaptive control system is designed, using fuzzy neural networks controller and adopting the learning arithmetic of modified diversified foot pace. The learning signal is offered by the neural network identification. Utilizing nonlinear mapping-capability of neural network can solve these problems caused by the dynamic characteristic well. Simulation results show, FNNC has improved the performance of original system, and it is better than normal PID controller in adaptability and robust.