Data centre loads have a high power density, strong power electronic coupling and rapid fluctuations. Voltage sags, harmonic distortion, three-phase imbalance, frequency drift and transient impacts can all be converted into service interruption risks by UPS, PDUs, switching power supplies and cooling systems. To address the problems of a high false alarm rate, insufficient offline diagnostic timeliness, and a lack of multi-indicator collaborative judgment in traditional threshold alarms, this paper introduces an online power quality anomaly identification and hierarchical alarm mechanism for data center power distribution links. Sliding window sampling is used to extract features from the following time-series data: root mean square voltage, total harmonic distortion rate, negative sequence voltage imbalance, crest factor and load mutation rate. A three-level hierarchy of identification is used: "basic threshold gating - multivariate anomaly scoring - alarm confidence confirmation", and alarm classification, suppression and regression strategies for closed-loop operation and maintenance are proposed. Based on the reanalysis of publicly available data centre power consumption statistics and published power quality identification benchmarks, it has been determined that this mechanism can raise the level of anomaly detection from a single point of failure to multi-evidence collaborative judgment, thus providing an implementable path for improving the reliability of data centre power supply management and capacity expansion, as well as proactive operation and maintenance. Supplementary robustness tests have been conducted by introducing noise, randomly missing data, multiple types of anomalies and detection latency, etc., to further validate the mechanism's suitability for edge deployment.

Xinyi Sun. Research on Online Identification and Alarm Mechanism for Abnormal Power Quality in Data Centers. International Journal of Engineering Technology and Construction (2026), Vol. 7, Issue 1: 8-17. https://doi.org/10.38007/IJETC.2026.070102.

[1] Shehabi A, Newkirk A, Smith S J, et al. 2024 united states data center energy usage report[J]. 2024.

[2] Chen, X.; Wang, X.; Colacelli, A.; Lee, M.; Xie, L. Electricity Demand and Grid Impacts of AI Data Centers: Challenges and Prospects. arXiv:2509.07218, 2025.

[3] Ginzburg-Ganz, E.; Lifshits, P.; Machlev, R.; Belikov, J.; Krieger, Z.; Levron, Y. Technical Challenges of AI Data Center Integration into Power Grids—A Survey. Energies, 2026, 19(1), 137. https://doi.org/10.3390/en19010137

[4] Sheng, Y.; Zhang, C.; Zhu, Z.; Xu, H.; Wen, J.; Wang, R.; Yang, J.; Wang, Q.; Bu, S. Power for AI Data Centers: Energy Demand, Grid Impacts, Challenges and Perspectives. Energies, 2026, 19(3), 722. https://doi.org/10.3390/en19030722

[5] Samanta, I. S.; Panda, S.; Rout, P. K.; Bajaj, M.; Piecha, M.; Blazek, V.; Prokop, L. A Comprehensive Review of Deep-Learning Applications to Power Quality Analysis. Energies, 2023, 16(11), 4406. https://doi.org/10.3390/en16114406

[6] Khetarpal, P.; Nagpal, N.; Al-Numay, M. S.; Siano, P.; Arya, Y.; Kassarwani, N. Power Quality Disturbances Detection and Classification Based on Deep Convolution Auto-Encoder Networks. IEEE Access, 2023, 11, 46026–46038. https://doi.org/10.1109/ACCESS.2023.3274732

[7] Perez-Anaya, E.; Jaen-Cuellar, A. Y.; Elvira-Ortiz, D. A.; Romero-Troncoso, R. J.; Saucedo-Dorantes, J. J. Methodology for the Detection and Classification of Power Quality Disturbances Using CWT and CNN. Energies, 2024, 17(4), 852. https://doi.org/10.3390/en17040852

[8] Gao, L.; Wang, J.; Zhang, M.; Zhang, S.; Wang, H.; Wang, Y. Classification Strategy for Power Quality Disturbances Based on Variational Mode Decomposition Algorithm and Improved Support Vector Machine. Processes, 2024, 12(6), 1084. https://doi.org/10.3390/pr12061084

[9] Bai, H.; Yao, R.; Zhang, W.; Zhong, Z.; Zou, H. Power Quality Disturbance Classification Strategy Based on Fast S-Transform and an Improved CNN-LSTM Hybrid Model. Processes, 2025, 13(3), 743. https://doi.org/10.3390/pr13030743

[10] Albalooshi, F. A.; Qader, M. R. Deep Learning Algorithm for Automatic Classification of Power Quality Disturbances. Applied Sciences, 2025, 15(3), 1442. https://doi.org/10.3390/app15031442