计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (5): 140-146.DOI: 10.3778/j.issn.1002-8331.1507-0215

• 模式识别与人工智能 • 上一篇    下一篇

全维状态观测器的演化设计

刘鹏皓,周永华   

  1. 广西大学 电气工程学院,南宁 530004
  • 出版日期:2017-03-01 发布日期:2017-03-03

Evolutionary design of full dimensional state observer

LIU Penghao, ZHOU Yonghua   

  1. College of Electrical Engineering, Guangxi University, Nanning 530004, China
  • Online:2017-03-01 Published:2017-03-03

摘要: 利用被控系统可以直接测量的输出量和输入量,作为观测器系统的输入信号,并使状态观测器的状态信号和被控系统的状态变量等价。以误差绝对值乘时间积分指标[ITAE]的倒数作为目标函数。利用智能体对环境的感知和反作用能力,将其与遗传算法的搜索方式相结合,提出了多智能体遗传算法来优化目标函数,进而设计出状态观测器的输出反馈矩阵。将多Agent系统与GA相结合的MAGA能够充分地挖掘GA中个体的并行性,从而在很大程度上提高算法的收敛速度和准确性。

关键词: 全维状态观测器, 多智能体, 遗传算法

Abstract: The output and input of the controlled system can be measured directly as the input signal of the observer system, and the state signal and the state variable of the controlled system are equivalent. With the countdown of the ITAE indicators as the objective function, the environmental perception and reaction ability of the agent is combined with genetic algorithm, and it puts forward the multi-agent genetic algorithm to optimize the objective function, and then designs the state observer’s output feedback matrix. The combination of multi-agent system and genetic algorithm can improve the parallelism of individual in GA, so as to greatly improve the convergence speed and accuracy of the algorithm.

Key words: full dimension state observer, multi-agent, genetic algorithm