摘要
为了控制循环流化床(CFB)锅炉的NOx排放量,以某热电厂300MW CFB锅炉测试数据为样本,应用支持向量机(SVM)建立NOx排放特性预测模型.针对SVM回归预测需要人为确定相关参数的不足,应用果蝇优化算法(FOA)优化SVM参数,采用不同工况下的样本数据检验FOA-SVM模型的预测性能,并将该模型的预测结果与粒子群算法(PSO)、遗传算法(GA)和万有引力搜索算法(GSA)优化的SVM模型预测结果进行了比较.结果表明:FOA-SVM模型的泛化能力较强,预测精度较高,训练时间较短,可以相对快速、准确地预测NOx排放质量浓度.
To control the NO~ emission from circulating fluidized bed （CFB） boilers, a model was estab lished based on test data of a 300 MW thermal power plant using support vector machine （SVM）. To over come the deficiency of SVM regression prediction in artificial determination of relevant parameters, the fruit fly optimization algorithm （FOA） was applied to optimize the SVM parameters. Prediction perform ance of the FOA-SVM model was then verified with sample data under different experimental conditions, of which the prediction results were compared with those optimized by particle swarm optimization （PSO）, genetic algorithm （GA） and gravitation search algorithm （GSA）. Results show that the FOA-SVM model has stronger genralization capability, higher prediction accuracy and shorter training time, which may therefore predict the mass concentration of NOX emission quickly and accurately.
出处
《动力工程学报》
CAS
CSCD
北大核心
2013年第4期267-271,共5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)
