基于EMD-LSTM模型的水轮机组实测摆度信号预测方法研究
Research on Prediction Method of Measured Swing Signal of Hydraulic Turbine Unit Based on EMD-LSTM Model
吴康平 1周建旭 1潘伟峰 2丁钶铖祺3
作者信息
- 1. 河海大学水利水电学院,江苏 南京 210098
- 2. 国网电力科学研究院南瑞集团有限公司,江苏 南京 211106
- 3. 河海大学电气与动力工程学院,江苏 南京 211100
- 折叠
摘要
水电机组的运行状态直接影响电站及电网的安全稳定,预测机组监测的振动信号有助于改善故障诊断的缺陷.为此,将经验模态分解(EMD)和神经网络模型相结合,提出一种基于 EMD-LSTM的水轮机组摆度信号预测模型,将该模型应用于国内某水电站的机组摆度信号预测中,并与 LSTM、GA-BP和 EMD-GABP模型预测结果进行比较.结果表明,该模型在机组摆度信号的预测方面表现出较高的精度,且优于其他模型.
Abstract
The operating condition of hydropower units is greatly related to the safety and stability of power stations and grids.The prediction of swing signals from unit monitoring can improve the defect of fault diagnosis.So,a combina-tion of empirical modal decomposition(EMD)and neural network model was used to put forward an EMD-LSTM-based model for predicting the swing signal of a hydropower station.The proposed model was applied to predict the swing sig-nal of a hydropower station in China,and the results were compared with those of LSTM,GA-BP and EMD-GABP mod-els.The results show that the model exhibits high accuracy in predicting the unit swing signal,outperforming other models.
关键词
水轮机组/摆度信号/经验模态分解/长短时记忆神经网络/预测精度Key words
hydraulic turbine sets/swing signal/empirical modal decomposition/long and short term memory neural networks/prediction accuracy引用本文复制引用
出版年
2024
NETL
NSTL
万方数据