Fault Diagnosis of Electric Submersible Pump System Based on Motor Current Signal Analysis and Deep Learning Method

Summary The petroleum industry relies heavily on electric submersible pump (ESP) systems, which can cause substantial economic losses if malfunctioning. However, over half of these malfunctions are due to submersible electric motor (SEM) failures. To address this issue, motor current signature analysis (MCSA) is used in this work to monitor the condition of the SEM. This allows for further fault diagnosis of the ESP system, ensuring that malfunctions are detected and addressed promptly. In this study, a portable high-speed electrical signal acquisition tool was used to capture the current signal at a 10-KHz rate to monitor the SEM condition. Afterward, the fast Fourier transform (FFT) method was used to transform time-domain data into the frequency domain for spectrum analysis. Next, the motor rotational speed, which is critical for identifying fault characteristics, was obtained from the current spectrum based on the deep learning method, and the fault components were extracted from the spectrum to indicate various faults. Finally, maintenance decisions were made based on the detected failure characteristics to prevent further damage and improve system reliability. Fusing MCSA with deep learning techniques significantly improved fault diagnosis accuracy compared with MCSA alone, demonstrating the robustness and effectiveness of the proposed method.