基于深度自编码器模型的汽轮机组异常状态检测研究
Research on Abnormal State Detection of Turbine Unit Based on Deep Auto Encoder Model
刘道庆 1臧润泽 1李鹏 1王振克 1程晨1
作者信息
- 1. 淮河能源电力集团有限责任公司,淮南 232087
- 折叠
摘要
文章提出一种提高汽轮机组安全性和效率的深度学习异常检测方法.使用长短时记忆(Long Short-Term Memory,LSTM)网络作为自编码器的网络层,分析学习时序数据特征,有效检测和提取异常状态数据.通过HAI数据集的训练测试,LSTM模型能够捕捉正常与异常状态的不同特征,识别潜在异常状态.对比实验和评估指标显示,提出的模型在异常检测任务上具有高准确率和召回率,并通过混淆矩阵验证了提出方法的有效性,进一步证明该方法在真实环境下对汽轮机组异常状态的检测具有良好的实用性和可靠性.
Abstract
This paper proposes a deep learning anomaly detection method to improve the safety and efficiency of a turbine unit.Using the Long Short-Term Memory(LSTM)network as the network layer of the auto-encoder,the analytical learning temporal data features are analyzed to effectively detect and extract abnormal state data.Through the training test of HAI dataset,the LSTM model captures the different features of normal and abnormal states to achieve the identification of potential abnormal states.Comparative experiments and evaluation metrics show that the proposed model has high accuracy and recall on the anomaly detection task,and the effectiveness of the proposed method is verified by the confusion matrix,which further proves that the method has good practicability and reliability in detecting the abnormal state of the turbine unit in a real environment.
关键词
长短时记忆网络/自编码器/异常状态检测/汽轮机Key words
Long Short-Term Memory(LSTM)/autoencoder/abnormal state detection/turbine引用本文复制引用
出版年
2024
NETL
NSTL
万方数据