Health Status Prediction of Hydraulic Automatic Grab Beam Based on CNN-LSTM
Aiming at the problem of degradation feature extraction and health state prediction of the automatic grab beam hydraulic system of large-scale hydropower station under monitoring data,a health prediction model was constructed based on convolutional neural network(CNN)and long short-term memory network(LSTM).Through the monitoring data,the degradation characteristics of hydrau-lic grab beams were extracted and predicted.The random forest was used to select the condition monitoring signal and the external envi-ronmental factors,the CNN was used to fully mine the spatiotemporal characteristics of the sensor data sequence,and the LSTM network was used to capture the information and dependence in the sequence data.In order to verify the effectiveness of the proposed model,the hydraulic automatic grab beam AMESim model based on real data was used for simulation verification.The results show that the deigned model achieves a higher prediction accuracy comparing with other traditional methods.
hydraulic automatic grab beamCNN-LSTMhealth status prediction
万方数据
维普
