Fault Diagnosis of Offshore Oilfield Cementing Equipment Based on DSD-CNN Deep Learning Model

Aiming at the complicated data characteristics such as strong noise, multi-sampling rate and dynamic correlation in the health monitoring of cementing equipment in offshore oilfield, a fault diagnosis method based on Denoising Synchronization Dynamic Convolutional Neural Network (DSD-CNN) deep learning model is proposed. Based on the traditional Convolutional Neural Network (CNN) model, Wavelet Transform (WT) technology is introduced to reduce data noises. Feature PreExtraction (FPE) technology is applied to the multi-sample rate data to synchronize the data sampling rate, and Long Short-Term Memory (LSTM) network is utilized to mine the internal temporal dynamic association of the data. DSD-CNN deep learning model extracts internal dynamic information from multi-source sensor data to achieve “end to end” efficient and accurate fault diagnosis. The application verification was carried out on the real data set of offshore oilfield cementing equipment operation. The results show that the proposed DSD-CNN deep learning model has better fault classification performance than the traditional CNN and LSTM models, and can provide important technical support for the safe production of offshore oil fields.