Physics-Informed Modeling of Temperature Effects in Rogowski Coils for Harmonic and Complex Signal Accuracy

Low-power instrument transformers (LPITs) are increasingly being deployed in power networks, either replacing or complementing traditional inductive instrument transformers (ITs). Although LPITs provide several advantages, they are also sensitive to a range of influencing factors. This work focuses on Rogowski coils (RCs) and investigates the impact of temperature on their performance. The analysis follows the temperature-accuracy testing procedure prescribed by current standards, with the results used to train a physics-informed deep learning model. The model is experimentally assessed for estimating the coil output over the considered temperature range and for harmonic components over a wide frequency spectrum. The proposed approach, validated on three commercial devices, enables the estimation of temperature-dependent output variations without the need for real-time thermal calibration or additional sensing hardware, thereby lowering system complexity and cost. By embedding physical knowledge into the data-driven framework, the methodology provides an interpretable model whose performance is evaluated at temperatures not used during training, within the experimentally tested range. The experimental findings confirm that combining domain knowledge with modern DL techniques can support the mitigation of one of the most critical error sources in RCs, enhancing their reliability for grid monitoring, protection, and control.