+Advanced Search

Fault Diagnosis Method of Chiller Sensor Deviation Based on CNN-GRU

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials

    To solve the problem of low error fault recognition rate of chiller sensors deviation,a sensor error fault diagnosis method based on Convolutional Neural Network(CNN)and Gated Recurrent Unit(GRU)fusion Net? work model(CNN-GRU)is proposed. This method utilizes GRU to remember the different time correlation of each sensor due to the different dynamic response characteristics of each sensor of the chiller,and overcomes the short? coming that CNN can only extract the real-time characteristics of time series in the fault diagnosis of sensor deviation of the chiller. First,the sensor real-time characteristics of time series are automatically extracted using CNN,and then GRU with a short and long-term memory capacity is used to realize the different time correlation of memory for the chiller sensor,so as to make full use of the feature information in the time series to characterize and model the data,and effectively improve the fault recognition rate of chiller sensor bias. Compared with CNN,PCA and autoen? coder methods,the experimental results show that the error fault identification rates of temperature and pressure sensors are more than 85% and 90%,respectively. The identification rate of deviation fault is more than 83%,which proves that the generalization ability of this method is good. The method has good symmetry for the fault recognition rates of the same sensor and the fault magnitudes are opposite to each other. The error fault identification rate of this method is higher than that of other methods,and especially for small error fault identification rate,it shows more ob? vious superiority.

    Cited by
Article Metrics
  • PDF:
  • HTML:
  • Abstract:
  • Cited by:
Get Citation
  • Received:
  • Revised:
  • Adopted:
  • Online: March 04,2022
  • Published: