Electricity generation is moving away from fossil fuels and towards renewable energy integration, which requires streamlined and cost-effective power distribution. Strategic modifications are essential to enhance the smart grid’s stability, safety, and efficiency, while minimizing the costs. One such modification is the replacement of traditional multi-wire conductors with sector-shaped multi-wire cables, which have superior conductive properties. To facilitate this transition, we explore the potential of contactless current measurements using fluxgate sensors and Machine Learning (ML) to predict symmetrical current flow from magnetic field data. This is necessary because mathematical formulations, such as the Biot-Savart Law and its variations, are inadequate for sector-shaped multi-wire cables. On the other hand, employing noninvasive sensors and ML reduces unwanted power losses and ensures good accuracy in current flow monitoring. Among the 18 ML regressors tested, the K-Nearest Neighbour Regressor predicted the amplitude with an error of 0.015 when tested on augmented data. The results of the experiments show that magnetic field reconstruction using ML is a worthy contender in our efforts to improve the smart power grid.
