摘要
针对汽轮机排汽焓参数难以直接测量的问题,提出了一种利用偏最小二乘法(PLS)和多种群遗传算法(MPGA)优化极限学习机(ELM)的汽轮机排汽焓预测模型。先将采集到的数据进行预处理,然后通过PLS将多维的输入变量降维成低维相互独立的变量,再利用MPGA对ELM的初始权值与阈值进行优化,最后使用优化后的模型进行训练。并将该模型预测结果与BP、ELM以及SVM等模型预测结果进行对比,结果表明MPGA-ELM具有更低的误差、更高的模型预测精度及更强的泛化能力。
Aiming at the problem that it is difficult to directly measure the exhaust enthalpy parameters of steam turbine, a prediction model of steam turbine exhaust enthalpy is proposed by using partial least squares(PLS) and multi population genetic algorithm(MPGA) to optimize extreme learning machine(ELM).Firstly, the collected data are preprocessed, and then the multi-dimensional input variables are reduced to low-dimensional independent variables by PLS.The initial weights and thresholds of ELM are optimized by MPGA.Finally, the optimized model is used for training.The results show that MPGA-ELM has lower error, higher prediction accuracy and stronger generalization ability than BP,ELM and SVM.
作者
齐继鹏
闫水保
冯灿
钱亿博
QI Ji-peng;YAN Shui-bao;FENG Can;QIAN Yi-bo(School of Mechanical and Power Engineering,Zhengzhou University,Zhengzhou,China,450001)
出处
《热能动力工程》
CAS
CSCD
北大核心
2021年第6期24-29,共6页
Journal of Engineering for Thermal Energy and Power
关键词
汽轮机排汽焓
偏最小二乘法
多种群遗传算法
极限学习机
steam turbine exhaust enthalpy
partial least square method
multi group genetic algorithm
limit learning machine
