基于CFNN的核蒸汽发生器水位控制

鉴于常规的PID控制存在控制对象参数变化时控制参数无法改变的不足,从而根据一个核蒸汽发生器(NSG)的简化数学模型,将一种补偿模糊神经网络(CFNN)用于NSG水位的控制。该网络由于引入了补偿神经元,使网络的容错性更好,系统更稳定。同时在神经网络学习算法中动态优化补偿模糊运算,使网络更适应,训练速度更快。仿真表明,该方法在装置负荷变化时比常规的PID控制方法超调量小,收敛速度快。该网络能在线调整参数,动态优化模糊规则,适于在线学习控制。该控制方法对NSG水位智能控制研究具有一定意义。

CFNN based water level control for nuclear steam generator

Because normal PID controller can't change its parameters according to the change of control object parameters.In this paper,the compensatory fuzzy neural network(CFNN) was used with a simplified model of nuclear steam generator(NSG) to design a NSG water level controller.Compensatory neurons which were introduced in the CFNN will make the control system improve the quality of fault tolerant and more stable.Meanwhile compensative fuzzy computation is optimized dynamically in the study algorithm of neural network,therefore the network is much more adaptive and the training speed is much faster.The results of simulation show that under this control method the system has smaller maximum overshoot and faster convergence speed than that of under normal PID control method.The CFNN can not only adjust parameters properly on line,but also can optimized relevant fuzzy reasoning in dynamic way,so it suit to be used on ling learning and control.The control method used in this paper is meaningful to the research of NSG water level intelligent control.