摘要
为了实现非线性、大时滞系统的良好控制，提出了一种新型的无辨识自适应预估控制算法。该算法将神经元结合到无辨识自适应控制律中，借鉴推导无辨识自适应控制参数自校正算式的基本思想建立约束条件，据此选择适当的权值取代原控制参数，并用加入动量项的改进δ算法取代该参数的校正计算式，提高控制参数的自校正能力。将该算法应用于600MW超临界机组直流锅炉的过热汽温控制，进行仿真研究，结果表明该算法的有效性，并具有良好的控制品质，较强的鲁棒性和自适应能力；且该算法对预估模型的精度要求不高，控制参数容易整定，易于工程实现。
For implementing excellent control to nonlinear systems with long time-delay, a novel adaptive predictive control algorithm of identification-free algorithm is proposed by combining a single neuron with the identification-free adaptive control algorithm. The restriction condition to select an appropriate weight to replace the original control parameter, is derived based on the basic principle applied to derive the self-adjusting equation of the identification-free control parameter. And the improved 8 weight-learning algorithm with momentum item replaces the self-adjusting equation in order to enhance its self-adjusting ability. A simulation for superheated steam temperature control of a supercritical once-through 600MW boiler using presented algorithm is carried out, and the results show the applicability, the excellent control performance and enhanced robustness and self-adaptability of the scheme in solving complicated system. The control parameters are easy to set because of undemanding precision to the predictive model, therefore, the scheme is easy for application.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2005年第2期103-108,共6页
Proceedings of the CSEE
基金
广西科学基金项目(桂科自0135065桂科基0448012)
关键词
单神经元无辨识自适应预估控制算法
数学模型
火电厂
锅炉
过热器
汽温控制
Thermal Power engineering
Predictive control
Identification-free
Adaptive control
Neuron
Superheated steam temperature
