基于多目标遗传算法的过热汽温建模与仿真  

Modeling and Simulation of Superheated Steam Temperature Based on Multi-objective Genetic Algorithm

作　　者：吴振龙 何婷[1] 王灵梅[2] 贾峰生 杨云凯 吴海曙 李东海[1] 韩磊 

机构地区：[1]清华大学热能系电力系统国家重点实验室,北京100084 [2]山西大学工程学院,山西太原030013 [3]国网山西省电力公司电力科学研究院,山西太原030001 [4]大同煤矿集团同达热电有限公司,山西大同037001

出　　处：《系统仿真学报》2017年第9期2081-2086,共6页Journal of System Simulation

基　　金：山西省煤基重点科技攻关项目(MD2014-07)

摘　　要：基于过热汽温的分布参数模型,采用高阶惯性传递函数进行逼近。为分析串级回路中导前区、惰性区模型阶次对辨识精度的影响和得到高阶惯性传递函数的动态参数,根据电厂实际运行的数据,采用多目标遗传算法对模型参数进行优化和仿真。通过分析仿真可知,阶次在合理的范围时阶次越高时帕累托前锋面越靠前,辨识具有更高的精确度。考虑实际需求与精度的要求,建立该机组过热汽温系统的串级模型。The high order inertial transfer functions were used to approximate the distribution parameter model. In order to compare the influence of the order of the leading region and the inertia object model on the accuracy and get the dynamic parameters, the multi-objective genetic algorithm was used to optimize the model parameters according to the actual operation data of the power plant. The higher the order, the Pareto front moves forward and recognizes the higher accuracy by the simulation when the orders are in a reasonable range. The reasonable superheated steam temperature system model was established considering the engineering and accuracy requirements.

关 键 词：过热汽温模型 串级系统 多目标遗传算法 PARETO最优解集 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]