Distribution Grid Monitoring Based on Widely Available Smart Plugs

TL;DR

This work addresses the challenge of observing distribution-grid state under increasing flexible loads by leveraging widely available smart plugs as high-frequency, low-cost measurement devices. It demonstrates that with firmware modifications, smart plugs can provide voltage measurements with substantially improved accuracy (SD around 0.27 V) and one-second cadence, enabling near real-time grid monitoring via an MQTT-driven data pipeline and time-series storage. A case study using an IEEE 37-bus grid shows that increasing the number of plugs steadily reduces the average monitoring error, validating the practicality of distributed, non-calibrated devices for DSOs. The approach offers a scalable, electrician-free deployment path for enhancing grid observability, while acknowledging privacy, local voltage drop, and single-phase limitations as areas for future work and extension to multi-phase networks.

Abstract

The growing popularity of e-mobility, heat pumps, and renewable generation such as photovoltaics is leading to scenarios which the distribution grid was not originally designed for. Moreover, parts of the distribution grid are only sparsely instrumented, leaving the distribution system operator unaware of possible bottlenecks resulting from the introduction of such loads and renewable generation. To overcome this lack of information, we propose the use of widely available smart home devices, such as smart plugs, for grid monitoring. We detail the aggregation and storage of smart plug measurements for distribution grid monitoring and examine the accuracy of the measurements. A case study shows how the average monitoring error in a distribution grid area decreases the more measurement devices are installed. Hence, simple smart plugs can help with distribution grid monitoring and provide valuable information to the DSO.

Figures (6)

• Figure 1: Smart plug measurements compared to voltage levels measured by a calibrated device
• Figure 2: Network infrastructure between the smart plug and the InfluxDB server
• Figure 3: Relative frequency histogram of the measurement error of smart plugs with the modified firmware (blue) and the unmodified firmware (orange).
• Figure 4: IEEE bus system 37. Nodes are colored based on the voltage monitoring error. Monitored voltages at white and yellow nodes are similar to the simulated voltages and monitored voltages at the red nodes differ more from the simulated voltages. In this instance, only one smart plug is installed at node 736 and the average monitoring error is about 2.87 V
• Figure 5: IEEE bus system 37. Nodes are colored based on the voltage monitoring error. Monitored voltages at white and yellow nodes are similar to the simulated voltages and monitored voltages at the red nodes differ more from the simulated voltages. In this example, eight smart plugs are installed at nodes 736, 706, 709, 711, 742, 722, 725 and 738. The average monitoring error is about 1.04 V and the coloring is consistent with Figure \ref{['fig:bus_system_37']}