基于改进分布式极限学习机的电站锅炉NOx排放预测算法

摘要

An improved distributed extreme learning machine was proposed to model the NOx emission characteristics of power station boiler.The introduction of distributed type and ridge re-gression theory improved the generalization performance and prediction accuracy of the limit learning algorithm.An improved MapReduce programming framework was adopted to carry out the parallelization of the the proposed algorithm so as to enhance its ability of dealing with mas-sive data.The real operation data provided by a 660MW power station boiler was analyzed and tested on Hadoop cluster.Results show that the proposed model has a better fitting and predic-tive ability for NOx emission,and the proposed algorithm has excellent parallel performance.%提出了一种改进的分布式极限学习机的电站锅炉NOx排放特性建模方法.引入分布式和岭回归理论,提升了极限学习机预测算法的泛化性能和预测准确率.采用改进的MapReduce编程框架对提出的算法模型进行并行化改进,提高其处理大数据的能力.选用某660 MW电站锅炉提供的真实运行数据进行分析,并在Hadoop集群上进行实验,结果表明该模型对NOx排放有着较好的拟合和预测能力,且提出的算法具有优异的并行性能.