全文下载:
202506014.pdf
文章编号: 1672-6987(2025)06-0111-07 DOI: 10.16351/j.1672-6987.2025.06.014
李泽, 张伟*(青岛科技大学 自动化与电子工程学院, 山东 青岛 266061)
摘要: 汽轮机组发电功率受主蒸汽的温度、压力和机组抽气、排气等的影响,在运行过程中易产生波动,导致电力输出突然过剩或减少。本文针对提高汽轮机组发电功率预测准确度的目的,提出一种基于CEEMDAN-LSTM-SVM的汽轮机组功率预测模型。首先使用完全自适应噪声集合经验模态分解方法(complete EEMD with adaptive noise,CEEMDAN)对功率序列进行解耦,将其分解成本征模式函数(IMF)分量和残余(RES)分量,然后对各分量分别进行长短期记忆网络(long short-term memory,LSTM)和支持向量机(support vector machine,SVM)预测,最后将预测数据采用拉格朗日乘子法进行最优权重相加得出最终结果。经对比实验,该模型的平均绝对误差(MAE)、平均绝对百分比误差(MAPE)、均方根误差(RMSE)更低。实验结果表明,该模型可对发电功率进行精确预测,有助于辅助汽轮机组运行调控。
关键词: 汽轮机组发电功率; 完全自适应噪声集合经验模态分解方法; 长短期记忆网络; 支持向量机; 时间序列预测
中图分类号: TP 181 文献标志码: A
引用格式: 李泽, 张伟. 基于CEEMDAN-LSTM-SVM的汽轮机组发电功率预测[J]. 青岛科技大学学报(自然科学版), 2025, 46(6): 111-117.
LI Ze, ZHANG Wei. Prediction of steam turbine generating power based on CEEMDAN-LSTM-SVM[J]. Journal of Qingdao University of Science and Technology(Natural Science Edition), 2025, 46(6): -.
Prediction of Steam Turbine Generating Power Based on CEEMDAN-LSTM-SVM
LI Ze, ZHANG Wei(College of Automation and Electronic Engineering,Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract: The generating power of the steam turbine unit is affected by the temperature and pressure of the main steam and the extraction and exhaust of the unit, which is easy to fluctuate during operation, resulting in sudden excess or reduction of power output. In order to improve the accuracy of power prediction for steam turbine units, this paper proposes a power prediction model for steam turbine units based on CEEMDAN-LSTM-SVM. First, the power series is decoupled using the fully adaptive noise set empirical mode decomposition method (Complete EEMD with Adaptive Noise,CEEMDAN), and the cost eigenmode function (IMF) component and residual (RES) component are decomposed, and then the short and long term memory network (Long Short-Term Memory,LSTM) and support vector machine (Support Vector Machine,SVM) prediction are carried out for each component respectively. Finally, the prediction data is added with the optimal weight using the Lagrange multiplier method to obtain the final result. Finally, the final results are obtained by adding the weights of the prediction data using the Lagrange multiplier method. Through comparative experiments, the mean absolute error (MAE), mean absolute percentage error (MAPE) and root mean square error (RMSE) of the model are lower. The experimental results show that the model can accurately predict the generating power, and is helpful to the operation regulation of steam turbine units.
Key words: generating power of steam turbine unit; complete EEMD with adaptive noise; long short-term memory; support vector machine; time series prediction
收稿日期: 2024-12-16
基金项目: 国家自然科学基金项目(61971253).
作者简介: 李 泽(1996—),男,硕士研究生. * 通信联系人.