K-聚类的模糊神经网络对DO的控制研究

摘要

Using a fuzzy Radial Basis Function(RBF)neural network based on K-clustering algorithm controls the concen-tration of quality of the dissolved oxygen(do)in the sewage treatment. This method combines fuzzy control reasoning ability and neural network learning ability characteristic. Fuzzy control, RBF neural network and K-clustering learning algorithm are applied in order to adjust subjection function on-line, optimize control rules. By the step input simulation analysis, the results show that fuzzy neural network controller based on the RBF has a good dynamic performance, strong robustness and anti-interference ability, make it fast and accurately to achieve the desired level.%运用一种基于K-聚类算法的模糊径向基函数（RBF）神经网络对污水处理中的溶解氧质量浓度进行控制，该方法结合了模糊控制的推理能力强与神经网络学习能力强的特点，将模糊控制、RBF神经网络以及K-聚类学习算法相结合以在线调整隶属函数，优化控制规则。通过对阶跃输入仿真分析，其结果表明基于RBF的模糊神经网络控制器具有良好的动态性能、较强的鲁棒性和抗干扰能力，使其快速、准确地达到期望水平。