基于粒子群算法的1000MW火电机组模型辨识

摘要

对于火电厂热工控制系统,建立精确的热工过程模型是保证控制质量的基础.将传统的阶跃响应法用于热工过程模型辨识,由于现场数据的不规则性使得阶跃响应法的经验结果精度不高.针对火电厂热工过程对象的特点及传统模型辨识的缺陷,将粒子群算法用于火电厂热工过程模型的辨识,以1000MW超超临界机组的风煤比作为控制对象,对空预器进口氧量变化系统进行系统辨识,辨识曲线能够很好的反应实际输出曲线,证明了该方法的有效性和可靠性,相比传统的辨识方法,将粒子群算法用于模型辨识提高了辨识的精确性与快速性.%For thermal power plant control system,to establish an accurate model of thermal processes is the basis of ensuring quality control.Because of the irregularities of the field data,we can not get accurate results through traditional step response method.Directing against features of thermodynamic process in thermal power plants and defects of traditional method for identification through model,a particle swarm algorithm was used in the thermodynamic process identification through model.Through identification of excess oxygen amount of boiler for a 1000MW unit,which takes air-coal ratio as the controlled object,the effectiveness and reliability of the said method were proven.Compared with traditional identification methods,the particle swarm algorithm improves the accuracy and rapidity.