Investigation on magnetostrictive behaviour of a converter transformer influenced by dominant harmonics: A FEM and ANN based approach

The introduction of harmonics to the HVDC system is imminent, which affects the components of the power system from varied perspectives. The present work emphasises analysing the behaviour of the converter transformer concerning the harmonics in the input. The process involves designing a stepped configuration of a 240-MVA converter transformer using the finite element method (FEM). The designed model is analysed as per its structural dynamics to obtain its vibrational response for different ranges over the frequency spectrum. Furthermore, the module of acoustics and thermal analysis is coupled to the vibrational model. The coupled model serves as an efficient tool in analysing the impact of different orders of harmonics on the vibrational, acoustic and thermal performance of the core of the converter transformer. The computational effort taken in solving the coupled field problem is simplified by designing a neural network model and the prediction efficacy of the network is enhanced by using a Hybrid of Particle Swarm Optimisation and Gravitational Search Algorithm (HPSOGWO) as tuner of the internal parameters of the neural network. The significance of the designed model is also examined against other well-versed hybrid algorithms in the literature in terms of various statistical indices.

Automated summary

This study focuses on the impact of harmonics on the input behavior of a converter transformer, designed a 240 MVA converter transformer, analyzed and tested vibrational response using the finite element method, and coupled with acoustic and thermal analysis modules. By optimizing internal parameters using a neural network model and the HPSOGWO algorithm, computational effort for solving the coupled field problem is simplified, and the designed model is compared with other hybrid algorithms in terms of various statistical indices.