摘要
针对一类随机NARMAX模型,分析了其可采用PID控制的约束条件。提出采用辅助模型的可克服算法病态的遗忘因子递推最小二乘算法对被控对象进行参数估计,利用动态切平面逼近的预测算法对系统输出进行预测,基于一具有预测控制性能的增量型预测滤波PID控制算法,根据可克服算法病态的直接极小化指标函数自适应控制算法和Robbins-Monro算法,给出了具有在线修正PID控制参数和加快PID控制参数收敛速度的随机NARMAX模型的自适应预测滤波PID控制算法。仿真研究表明:因给出的PID控制算法具有预测控制性能和在线修正参数性能,故系统具有较好的控制品质。
The constraint conditions being applicable to the stochastic muhivariable NARMAX model were analyzed. The parameter estimation of the controlled system was conducted by using the nonlinear muhivariable forgetting factor recursive least squares algorithm with solving ill-controlled of the auxiliary model, and the output prognosis of the controlled system was conducted by using the dynamic cutting horizontal approximating algorithm. Based on the incremental prediction filter decoupling PID control algorithm with the characterization of prediction- control, the self-tuning control algorithm of direct minimization index function with solving ill-controlled and the Robbins-Monro algorithm, a prognosis-filtering PID control algorithm with the characterizations of the on-line modifying parameter and the speeding the convergence of PID control parameter due to the index function containing the predicting values of the outputs was developed for the stochastic NARMAX model. The simulative results indicate that the system exhibits very good controlling characterization due to the developed PID control algorithm with the properties of predicting-controlling and on-line modifying parameter
出处
《北京联合大学学报》
CAS
2016年第4期41-47,共7页
Journal of Beijing Union University
关键词
自适应控制
预测控制
PID控制
参数估计
动态切平面逼近
随机NARMAX模型
Adaptive control
Predictive-control
PID control
Parameter estimation
Dynamic cutting horizontal approximating
Stochastic NARMAX model
