Power quality estimation helps utilities and industrial stakeholders assess electrical conditions at locations where no monitor is installed. This matters because distribution networks must accommodate modern power-electronic devices and distributed generation, which can both respond to and contribute to power quality disturbances. Harmonic distortion can overheat equipment or interfere with sensitive devices, while voltage dips can interrupt industrial processes. Since monitoring every network node is impractical, measurement and estimation need to work together.
For engineers and system integrators, the practical challenge is not simply collecting data. It is interpreting measurements accurately, understanding their limitations, and using them to estimate conditions elsewhere in the network. Two phenomena are especially important: harmonics and voltage dips.
Why power quality estimation is needed
Power quality measurements characterize voltage and current at monitored points, support compliance assessments, and help investigate disturbances. But measurement coverage is constrained by the cost of instruments and communications infrastructure. A limited set of monitors cannot directly reveal the power quality state at every bus.
State estimation addresses this gap. It uses available measurements and a representation of the network to infer conditions at locations without meters. This is an underdetermined problem: there may be more unknown network conditions than measured values. Measurement error and uncertainty in network parameters add to the challenge, so estimates should be interpreted with those limitations in view.
Estimating harmonic distortion across a network
Harmonics are sinusoidal components at integer multiples of the fundamental frequency. They distort the voltage or current waveform and can arise from power-electronic converters, non-linear appliances, and equipment with magnetic cores. Harmonic assessment commonly considers individual harmonic levels and total harmonic distortion (THD), which expresses the combined harmonic content relative to the fundamental component.
Why magnitude-only measurements can mislead
When harmonic currents from several devices or feeders combine, their phase angles affect the resulting total. The currents may partially cancel, meaning their combined magnitude can be lower than the arithmetic sum of their individual magnitudes. Estimating the total using only measured magnitudes and summation coefficients can therefore produce substantial error.
Measuring harmonic phase angles provides information about this summation effect. However, a complete measurement procedure must address more than capturing an angle at one instant: it also needs to specify how spectral components are extracted and how measurements are aggregated over time. Considering phase angles can improve the basis for estimating the combined harmonic current, while the measurement process itself introduces uncertainty that should be assessed.
Modelling sources and locating disturbances
Network analysis also depends on useful harmonic source models. A bottom-up stochastic approach can represent residential harmonic injections using factors such as household occupancy, appliance allocation, appliance operating behaviour, and measured harmonic current spectra. The model can incorporate possible changes in connected devices and can be checked against field measurements.
To identify likely disturbance locations, Bayesian inference can combine measured harmonic quantities with information about possible injections, changing grid conditions, and uncertainty in measurements and network parameters. Rather than treating one location as certainly responsible, the approach can produce a ranking of suspected sources. Network operators can use that ranking to guide troubleshooting and decide where further investigation or corrective measures may be appropriate.
Estimating voltage dips at unmonitored buses
A voltage dip is a temporary reduction in voltage magnitude below a specified threshold. Short circuits are a common cause, although events such as motor starting, transformer energizing, and load switching can also produce dips. Their key characteristics are residual voltage—the lowest RMS voltage during the event—and duration, measured from the voltage falling below the threshold until it recovers.
Dip measurement relies on RMS voltage values, so the calculation and detection method affect the reported event. In particular, dips with short durations or relatively shallow magnitudes can be significantly affected by detection uncertainty. Engineers should therefore consider the measurement algorithm when comparing results or assessing equipment response.
Bayesian inference can also support voltage dip state estimation. By combining measurements from monitored buses with information about pre-fault and faulted network conditions, the method estimates residual voltages at unmonitored buses. A component failure database can inform the prior distribution of likely fault locations, while measurement-device and component-parameter uncertainties are incorporated into the analysis. Dip duration can also be estimated using protection-system time settings and their uncertainties.
Planning measurements for useful estimates
Estimation performance depends partly on where and what the network measures. The number and location of measurement points, the quantities recorded, and the accuracy of instruments can all influence results. Sensitivity studies help determine how these choices affect performance and what measurement structure is needed for a specified estimation objective.
- For harmonic assessments: consider phase-angle information when evaluating current summation, and account for uncertainty in both measurements and network parameters.
- For disturbance investigation: use estimated source rankings to prioritize locations for follow-up rather than treating estimates as direct measurements.
- For voltage dips: assess RMS calculation and detection uncertainty, particularly for short or shallow events.
- For network planning: evaluate how measurement coverage affects estimates at unmonitored buses before deciding where additional monitoring may be useful.
Conclusion
Power quality estimation extends the value of a limited monitoring network, but its usefulness depends on sound measurement methods, realistic source and grid models, and explicit treatment of uncertainty. Harmonic phase angles help explain how currents combine, while statistical approaches can rank potential disturbance locations. For voltage dips, RMS detection accuracy and knowledge of fault conditions shape estimates of residual voltage and duration. Together, measurement and estimation give network operators and industrial decision-makers a more complete basis for assessing disturbances and directing further investigation.