湖南电力 ›› 2025, Vol. 45 ›› Issue (3): 129-134.doi: 10.3969/j.issn.1008-0198.2025.03.018

• 电力人工智能与数字化 • 上一篇    下一篇

基于DE-MLRF的汽轮机负荷偏差识别方法

文雯1, 黄煜杰2, 唐明珠2, 邓黎明1, 罗翠娥1, 姜鑫1   

  1. 1.湖南大唐先一科技有限公司,湖南 长沙 410007;
    2.长沙理工大学能源与动力工程学院,湖南 长沙 410114
  • 收稿日期:2025-01-13 修回日期:2025-01-24 出版日期:2025-06-25 发布日期:2025-07-02
  • 通信作者: 唐明珠(1983),男,博士,教授,主要从事电力设备智能运维研究工作。
  • 作者简介:文雯(1989),女,硕士,主要从事电力设备智能运检研究工作。
  • 基金资助:
    国家自然科学基金项目(62173050); 湖南省自然科学基金项目(2023JJ30051); 湖南省重大科技创新平台项目(2024JC1003)

A Identifying Method for Turbine Load Deviation Based on DE-MLRF

WEN Wen1, HUANG Yujie2, TANG Mingzhu2, DENG Liming1, LUO Cui′e1, JiANG Xin1   

  1. 1. Hunan Datang Xianyi Technology Co., Ltd., Changsha 410007, China;
    2. School of Energy and Power Engineering, Changsha University of Science & Technology,Changsha 410114, China
  • Received:2025-01-13 Revised:2025-01-24 Online:2025-06-25 Published:2025-07-02

摘要: 针对汽轮机变工况运行存在负荷偏差的问题,提出一种基于差分进化算法(differential evolution,DE)和多标签随机森林(multi-label random forest,MLRF)结合的汽轮机负荷偏差原因分类模型。通过斯皮尔曼(Spearman)相关性系数,分析影响汽轮机负荷出力的相关联变量;采用DE算法优化MLRF模型参数,建立基于DE-MLRF的汽轮机负荷偏差原因分类模型。结合某660 MW汽轮机实际运行数据进行实验验证,结果表明,与其他7种算法相比,DE算法优化的MLRF模型误报率(1.902 4%)、漏报率(1.854 1%)最低,可为汽轮机负荷偏差原因定位提供决策支持。

关键词: 汽轮机, 负荷偏差, 差分进化算法, 多标签随机森林

Abstract: Aiming at the problem of load deviation in turbine variable condition operation, a classification model for the reason of turbine load deviation is proposed based on the combination of differential evolution (DE) algorithm and multi-label random forest (MLRF). The associated variables affecting the turbine load output are analyzed by Spearman correlation coefficient. The DE algorithm is used to optimize the parameters of the MLRF model, and the DE-MLRF-based classification model of the causes of turbine load deviation is established. The experimental validation is carried out by combining the actual operation data of a 660 MW turbine, and the results show that the MLRF model optimized by DE algorithm achieves the lowest false alarm rate and missing alarm rate, which are 1.902 4% and 1.854 1% respectively, compared with MLRF models optimized by the other seven algorithms. The DE-MLRF-based model for classifying the causes of turbine load deviation can provide decision support for locating the causes of turbine load deviation.

Key words: turbine, load deviation, differential evolution (DE) algorithm, multi-label random forest (MLRF)

中图分类号: