一种规则简化的模糊神经网络控制器

构造了一种实时模糊神经网络控制器，为解决模糊规则组合爆炸问题提供一个新方案。控制器基于T-S模糊模型，由前后件分离的网络结构实现。前件参数通过移动小论域法创建，每个变量仅在工作小论域上生成两个模糊子集，有效减少模糊规则，增强实时性；后件参数通过有ki,kp,kd修正因子的BP改进算法在线更改，控制意义明确，确保系统动态性能。仿真结果证实该控制器实时性好，控制性能优，鲁棒性强。

A Fuzzy Neural Network Controller with Simplified Rules

A real-time fuzzy neural network controller is proposed in this paper. It gives a new approach to solve combination explosion of fuzzy rules. Network structure of controller based on T-S fuzzy model comprises premise part and consequence part. A small dynamic universe method is employed to generate premise parameters, which efficiently reduces numbers of fuzzy rules and improves the real-time performance of algorithm by only dividing two fuzzy subsets for every input in the operating universe. An improved BP algorithm with ki, kp, kd coefficient modifies consequence parameters, which clarifies meaning of parameters and ensures the transient response. Simulation results demonstrate that the controller has good real-time performance, control quality and robustness.