摘要
针对基于群体智能算法的热工过程参数辨识存在随机性强、收敛慢、耗时长的不足,提出了一种快速辨识方法。在仿真验证辨识目标函数随各个参数变化呈单峰性后,将黄金分割法与坐标轮换法相结合,形成逐维黄金分割法,并通过选取最优初始点和提高寻优精度等方法对其进行改进;基于现场数据分别采用粒子群优化算法、逐维黄金分割法及其改进算法对风煤比-过氧量模型的过程参数进行了辨识比较。结果表明,改进后的逐维黄金分割法在快速性、精确性上明显优于粒子群算法和逐维黄金分割法,其更适合于热工过程参数的在线辨识,从而为热工控制系统调节参数的在线快速优化提供条件。
The thermal process parameters identification by swarm intelligence algorithm has disadvantages of strong randomness,slow convergence and time-consuming.Against this problem,this paper proposes a quick identification method.First of all,simulation was carried out to verify that the identification problem＆#39;s target function shows unimodal characteristics to each parameter.Then,combing the golden section method with coordinate alternation method,the author proposed the dimension-by-dimension golden section method,and improved it by selecting the optimal initial point and enhancing the optimization accuracy.Fi-nally,on the basis of the field tests data,the particle swarm optimization （PSO）algorithm,dimension-by-dimension golden section method and its improved method were applied to identify the process parameters of air/coal ratio and excess oxygen content model.The results show that,the improved dimension-by-di-mension golden section method is superior to the other two methods in quickness and accuracy,which is more suitable for online identification of the thermal process parameters,so it provides the conditions for adj usting parameters＆#39;online optimization in thermal control system.
出处
《热力发电》
CAS
北大核心
2015年第10期68-71,76,共5页
Thermal Power Generation
关键词
热工过程
参数辨识
逐维黄金分割法
坐标轮换法
粒子群优化算法
风煤比
过氧量
thermal process
parameter identification
dimension-by-dimension golden section method
coor-dinate alternation method
particle swarm algorithm
air to coal ratio
excess oxygen content
