发酵过程自学习模糊神经元控制器的设计

对发酵过程采用常规的控制方式,其控制效果不好,甚至难以实现稳定控制。高级控制算法一般需要大量的先验知识,对过程精确模型依赖较大。而模糊逻辑控制技术一般用来控制那些具有模糊性、不确定性、高阶、大滞后等难以用精确的数学模型来描述的对象;神经网络具有学习、记忆等能力。采用自组织计数传播网络(CPN)作为框架,结合改进的模糊控制算法,实现对发酵过程的模糊神经元控制。该方法有能力自组织、自学习发酵过程所需的控制知识,规则库初始为空,逐渐地被自构造,来满足预先设定的性能要求。通过对发酵过程控制的仿真研究,表明该方法能够实现自学习的能力。

Design of Self-learning Fuzzy Neural Controller for Fermentation Process

In fermentation process, when routine control algorithm has been used, the effect of control is bad. Even it is difficult to realize a stable control. Advance control algorithm usually needs much more prior knowledge and depends on the accuracy model of process. However, fuzzy logic control technology is applied to control the plants having fuzzy, uncertainty, high-order, heavy lag without accurate mathematics model. The neural network has the advantage of self-learning, memory ability, fault-tolerant and parallel processing etc. The count propagation network (CPN) was taken as framework, combining an improved fuzzy control algorithm, to realize the fuzzy-neural control of fermentation process. The method has the ability of self-organizing and self-learning the control knowledge which is needed for fermentation process. The rule-base initially is empty, and is self-constructed gradually, to meet the performance index. Simulation results prove that the method can realize the ability of self-learning.