Dynamic genetic algorithm in model for vibrations of power transformer windings

Abstract

Power transformers are vital components of the power system. These electrical devices can experience multiple types of faults within them. Mechanical defects are relatively difficult to detect since they occur inside the structure of the power transformer. Magnetostriction can create this type of faults in magnetic core and electrodynamic forces can cause them in windings. Diagnostic methods mainly use vibration information from the surface of transformer. This is an indirect approach and causes additional errors since vibrations need to propagate transformer structure before sensors receive them. However, it is possible to calculate the mechanical situation in windings by modelling this part of the transformer as a mass and spring system with forces acting upon it and applying dynamic genetic algorithm. This paper describes a model created in COMSOL and Matlab software and a case study for a segment of windings displays the acquired results.