Power quality analysis and energy consumption monitoring across hospitals in Cameroon

Abstract

The quality of electrical supply is the main cause of premature failure of biomedical equipment in Cameroon, where nearly 95% of hospital devices are imported and designed to operate under stable mains conditions. This study, based on real hospital data, utilises data collected by CURES (University Centre for Energy Research in Healthcare) in four Cameroonian hospitals. The data collection yielded 3,474 ten-minute intervals. A model-free algorithm enabled, in turn: the profiling of regulatory compliance against the EN 50160 and IEEE 519-2014 standards; the detection of anomalies via SPE and Hotelling’s T2 statistics derived from PCA; the localization of faults through residual structuring; unsupervised classification using the K-means method validated by the elbow and silhouette criteria; and finally, the modelling of temporal dependence via a robust vector autoregressive model with parameter estimation based on the M-estimator. Four disturbance profiles were identified. A Spearman’s correlation of ρ = −1.00 between anomaly rates and compliance scores indicates the internal consistency between the compliance scoring and the anomaly detection outputs. With a frequency compliance rate of 28.9% and a score of 47.8%, Djoum Hospital is the most critical site, while Obala achieved the highest score (72.6%). These results highlight the urgent need to deploy reactive compensation and frequency stabilisation systems in the rural hospital facilities included in this study.