论文部分内容阅读
自适应控制系统往往结合神经网络技术和模糊理论来实现规则节点和隶属度函数调整。但是这种系统的运行过程往往是顺序的,自适应过程慢。此外,在网络结构中往往存在冗余节点,加大了计算量,降低了控制反应速度。针对以上问题本文设计1个新的模糊神经网络控制系统(FNCC),FNCC在结构学习中引入了减少规则节点的操作,降低了由于过量计算所带来的时间滞后。同时,此系统的参数学习与网络结构学习同步进行,降低了由于顺序操作所带来的时间滞后。通过研究得出FNCC具有以下特点:(1)无需预知系统的模型,(2)无限制的结构设计。在研究中我们将此系统应用到一非线性系统上,通过仿真结果来验证FNNC的可行性和准确性。
Adaptive control system often combined with neural network technology and fuzzy theory to achieve the rules node and membership function adjustment. However, the operation of such systems is often sequential, and the adaptation process is slow. In addition, redundant nodes often exist in the network structure, increasing the amount of computation and reducing the control response speed. Aiming at the above problems, a new fuzzy neural network control system (FNCC) is designed in this paper. FNCC introduces the operation of reducing regular nodes in the structure learning and reduces the time lag caused by overcomputation. At the same time, the parameter learning of this system is synchronized with the learning of network structure, reducing the time lag caused by the sequential operation. Through the study, FNCC has the following characteristics: (1) no need to predict the system model, and (2) unlimited structure design. In the study, we apply this system to a nonlinear system, and verify the feasibility and accuracy of FNNC through simulation results.