TBD

This dataset presents a "Physics-Informed for Digital Twin" uniquely designed to simulate the electrical behavior of a typical household in Hyderabad, India.  Unlike standard datasets that assume a stable international voltage, this simulation is anchored to a verified high-voltage baseline of 253.53V and a drifting frequency of 49.96Hz. These specific values are directly derived from local smart meter validation data to ensure maximum realism. The core innovation is the use of a ZIP (Constant Impedance, Current, Power) physics engine. This system enforces strict electrical laws rather than relying on simple statistical patterns. Consequently, it ensures that appliance power consumption reacts physically to household electrical stress. For example, it correctly models how a resistive load's power drops quadratically during a voltage dip. This offers a level of fidelity and responsiveness that purely data-driven models cannot match.

The critical value of this archive lies in its ability to simulate rare and potentially hazardous household electrical events. It spans over one million data points at a ten-second resolution, injecting physics-accurate anomalies. These include severe voltage drops to 180V and realistic sensor noise, which are explicitly coded into the simulation script. All these events are paired with perfect ground-truth labels for seven parameters: Time, Voltage, Current, Power, Energy, Frequency, and Power Factor.