Implementation of Current Transformer Algorithm Based on Machine Learning

With the development of the construction of "Smart Grids", new requirements have been raised for the efficiency and accuracy of current measurement. Nowadays, current measurement system based on giant magnetoresistance effect (GMR) becomes a new research direction in related fields. The working principle of this measurement system is obtaining the measured current information indirectly by analyzing the magnetic field data, which is collected by a series of GMR magnetic field sensor array around the wire. Essentially, it is an inverse problem from magnetic field to current. At present, optimization algorithm is mainly used for this kind of inverse calculation, which, however, is difficult to balance the efficiency and accuracy of the algorithm. Thus, we propose the idea of realizing the inverse calculation by using machine learning. Based on a specific kind of circular sensor structure, we propose a neural network-based inverse calculation algorithm and verifies the feasibility of this algorithm.