摘要
关于难以建模的非线性系统的控制问题,提出具有辅助变量的全格式动态线性化方法逼近非线性系统模型,基于其构建系统的预测模型,给出采用直接极小化指标函数自适应优化算法的参数估计算法,在扩张状态观测器中引入控制输入的微分项,并将控制输入和其微分的系数改进为关于观测状态的函数,因其未知,使用RBF神经网络逼近,利用非线性递推最小二乘法同时优化RBF神经网络参数和自抗扰控制器参数,综上研究提出在线优化参数的无模型预测神经网络自抗扰控制算法。仿真研究验证了上述研究的合理性和有效性,系统响应精度高。
Regarding the control issues of complex nonlinear systems that are difficult to model,a comprehensive dynamic linearization method incorporating auxiliary variables is proposed to approximate the nonlinear system model.Parametric estimation algorithm is obtained by using the adaptive optimization algorithm for direct minimization of index function.Differential term of control input is introduced into the extended state observer,and the control input and the factor of its differential term are transformed into function of observer status.Due to its unknown nature,an RBF neural network is employed for approximation.The nonlinear recursive least squares method is utilized to simultaneously optimize the parameters of the RBF neural network and the active disturbance rejection controller.In summary,this research proposes a model-free predictive neural network active disturbance rejection control algorithm with online parameter optimization.Simulation studies have validated the rationality and effectiveness of the proposed approach,demonstrating high system response accuracy.
作者
侯小秋
Hou Xiaoqiu(School of Electronics and Controlling Engineering,Heilongjiang University of Science and Technology,Harbin City,Heilongjiang Province 150022)
出处
《黄河科技学院学报》
2024年第8期12-18,共7页
Journal of Huanghe S&T College
关键词
自抗扰控制
神经网络控制
无模型自适应控制
预测控制
非线性系统
直接极小化指标函数自适应优化算法
非线性递推最小二乘法
在线优化参数
active disturbance rejection control
neural network control
model-free adaptive control
predicative-control
nonlinear system
adaptive optimization algorithm for direct minimization of index function
nonlinearity recursive least squares method
parameters optimization on-line
