An open rack vaporizer is a facility that vaporizes liquefied natural gas using sea water. When a vaporization efficiency of the open rack vaporizer decreases, liquefied natural gas can leak, which can cause great damage to the facility. Operators have to monitor the operation state of the facility in real-time to prevent the accident. However, operators have visited the site and have checked the state by looking at the value of sensors installed in the open rack vaporizer through indicators. For the safe operation of the open rack vaporizer, a monitoring system is needed to monitor the operation state of the open rack vaporizer in real-time without the need for operators to visit the site. In this paper, we developed a long term evolution based monitoring system to monitor the operation state of the open rack vaporizer. The developed system can monitor the real-time operation state of the open rack vaporizer at a control center far from the facility. For the system development, data transmission infrastructure using long term evolution was built. Afterwards a software was developed to monitor the operation state of the open rack vaporizer in real-time using the transmitted data. Finally, performance evaluation was conducted to confirm that the developed system operated successfully without data transmission delay or data missing.

We propose a phase-aware complex-spectrogram autoencoder (AE) for preprocessing raw vibration signals of rotating electrical machines. The AE reconstructs normal components and separates fault components as residuals, guided by an input-phasor phase-orthogonality regularization that defines parallel/orthogonal residuals with respect to the local signal phase. We use a U-Net-based AE with a mask-bias head to refine local magnitude and phase. Decisions are based on residual features—magnitude/shape, frequency distribution, and projections onto the normal manifold. Using the AI Hub open dataset from field ventilation motors, we evaluate eight representative motor cases (2.2–5.5 kW: misalignment, unbalance, bearing fault, belt looseness). The preprocessing yielded clear residual patterns (low-frequency floor rise, resonance-band peaks, harmonic-neighbor spikes), and achieved an area under the receiver operating characteristic curve (ROC-AUC) = 0.998–1.000 across eight cases, with strong leave-one-file-out generalization and good calibration (expected calibration error (ECE) ≤ 0.023). The results indicate that learning to remove normal structure while enforcing phase consistency provides an unsupervised front-end that enhances fault evidence while preserving interpretability on field data.