实用NARMAX-L2模型的神经网络自校正控制器  

Self-Tuning Controller of Practical NARMAX-L2 Model Neural Network

作　　者：侯小秋[1] 李丽华[1] Hou Xiaoqiu;Li Lihua(School of Electronics and Controlling Engineering,Heilongjiang University of Science and Technology,Haerbin City,Helongjiang Province150022)

机构地区：[1]黑龙江科技大学电气与控制学院,黑龙江哈尔滨150022

出　　处：《黄河科技学院学报》2023年第8期74-78,共5页Journal of Huanghe S&T College

摘　　要：针对实用随机非线性NARMAX模型,参照改进的NARMA-L2模型,提出实用随机NARMAX-L2模型,采用BP神经网络逼近该模型,依据直接极小化指标函数自适应优化算法估计随机干扰模型的参数,并对BP神经网络的连接权进行学习,根据广义控制目标函数提出神经网络显式自校正控制器,仿真研究验证了上述研究的有效性,系统具有理想的控制效果。With regard to the practical nonlinear NARMAX model,referring to modified NARMAX-L2 model,and using BP neural network to approach to the model.Parameter of stochastic excitation model was estimated by adaptive optimization algorithm for direct parameter estimation,the and connecting weight of BP neural network were learned.An explicit self-tuning controller was developed by using generalized control object function.Simulation result indicates that the controller has high effectiveness,and the system has excellent control performance.

关 键 词：神经网络控制 自校正控制:非线性系统 广义目标函数 直接极小化指标函数的自适应优化算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]