摘要
针对目前国内外对于冷水机组传感器偏差故障检测效果不理想的问题,结合长短期记忆网络(LSTM)适用于处理高维、强耦合、高度时间相关性数据的特点,该文提出一种基于改进LSTM的深度学习方法,用于冷水机组传感器偏差故障检测。现场采集风冷冷水机组传感器数据,用于训练改进的LSTM。通过实验分析得出,不同传感器检测效率不同。将该文所提方法的检测结果与自动编码器(Auto encoder)、主元分析法(PCA)、标准的LSTM三种方法的检测结果进行比较,得出该文所提方法在冷水机组传感器偏差故障检测中检测效率明显优于其他三种方法;并且针对同一传感器相同大小、不同正负的偏差故障,所提方法的检测效率具有更好的对称性。最后证明该文所提的改进LSTM方法具有良好的泛化性。
At present,the sensor deviation fault detection of the water-cooled chiller is unexpected all over the world.Long short-term memory(LSTM)is suitable for processing highdimensional,high-coupling,high-time correlation data.Taking into account the characteristics of LSTM,this paper proposed a deep learning method based on improved LSTM for the sensor deviation fault detection of the water-cooled chiller.The paper collected sensor data of the watercooled chiller on site to train the improved LSTM network.It could be know through simulation experiments that the detection efficiency of different sensors is different,and the results were compared with the results of three other methods:auto encoder,principal component analysis(PCA)and standard LSTM.Finally,the detection efficiency of the method proposed in this paper is significantly better than the other three methods in sensor deviation fault detection of the watercooled chiller.Furthermore,the detection efficiency of the proposed method has better symmetry for positive and negative fault levels with same absolute magnitude.Finally,it was proved that the proposed network has good generalization ability.
作者
李冬辉
尹海燕
郑博文
刘玲玲
Li Donghui;Yin Haiyan;Zheng Bowen;Liu Lingling(School of Electrical and Information Engineering Tianjin University Tianjin 300072 China)
出处
《电工技术学报》
EI
CSCD
北大核心
2019年第11期2324-2332,共9页
Transactions of China Electrotechnical Society
关键词
长短期记忆网络
深度学习
冷水机组
传感器
故障检测
Long short-term memory
deep learning
water-cooled chiller
sensor
fault detection
