基于万有引力搜索算法的电厂锅炉NO_x排放模型的参数优化  

Parameter Optimization for NO_x Emission Model of Power Plant Boilers Based on Gravitational Search Algorithm

作　　者：牛培峰[1,2] 肖兴军[1] 李国强[1] 马云飞[1] 陈贵林[1,2] 张先臣[1,2] 

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,秦皇岛066004 [2]国家冷轧板带装备及工艺工程技术研究中心,秦皇岛066004

出　　处：《动力工程学报》2013年第2期100-106,共7页Journal of Chinese Society of Power Engineering

基　　金：国家自然科学基金资助项目(60774028);河北省自然科学基金资助项目(F2010001318)

摘　　要：以某330MW煤粉汽包锅炉为测试对象,以支持向量回归机为基础,利用最小二乘支持向量机和一种新的寻优算法———万有引力搜索算法进行了综合建模及参数优化.结果表明:建立的模型较好地实现了对电厂锅炉NOx质量浓度排放量的预测;与遗传算法、蜂群算法和粒子群算法相比,万有引力搜索算法能更好地找到未知优化参数,使得所建模型具有更高的预测能力和泛化能力,从而有效控制燃煤电厂NOx的排放量.Taking a 330 MW coal fired boiler as an object of study, an optimized NOx emission model was established for the boiler using Least Squares Support Vector Regression （LSSVR） and Gravitational Search Algorithm （GSA）, with which the boiler NOx emission was successfully predicted. Results show that compared with Genetic Algorithm, Artificial Bee Colony and Particle Swarm Optimization, the GSA algorithm has a higher capability in parameter optimization, emission prediction and data generalization, which may therefore help to reduce NO, emission of power plants.

关 键 词：万有引力搜索算法 最小二乘支持向量机 燃烧优化 NOX排放 参数优化 

分 类 号：TK224.9[动力工程及工程热物理—动力机械及工程]