A comparative assessment of open-hardware and commercial energy meters for non-intrusive load monitoring

The accuracy, resolution, and spectral content of electrical measurement systems play a critical role in the reliability of energy analysis applications. This paper presents a comparative application-level assessment of open-hardware and commercial energy meters, examining how their metrological characteristics relate to performance in non-intrusive load monitoring (NILM) tasks within a common evaluation framework. Two open-hardware instruments, OpenZmeter (oZm) and Open Multi Power Meter (OMPM), are assessed against reference datasets acquired with commercial meters using identical energy disaggregation workflows implemented in NILMTK and widely adopted performance metrics. In addition to conventional electrical quantities, the effect of high-resolution harmonic measurements, available up to the 50th order in oZm, is explicitly analyzed. The results show that OMPM achieves very low Energy Assignment Error (0.006) and low Root Mean Square Error (14.782). At the same time, oZm provides application-level performance that is competitive with that of professional power-quality analyzers under the evaluated conditions. The inclusion of harmonic information increases the F1-score by 10–15% and reduces error metrics by more than 40% for loads with distinctive spectral signatures. A normalized GlobalScore, used here as an exploratory aggregate indicator, was computed for the subset of datasets with complete metric availability and revealed consistent comparative trends between measurement quality and NILM performance. Although cross-dataset differences in load composition, operating conditions, and acquisition settings limit strict one-to-one equivalence, the results support the use of open-hardware energy meters as reliable, reproducible, and cost-effective instrumentation for NILM-oriented energy monitoring and research