Software Approaches for Designing Electric Transformers

Abstract

In Korea, there are various transformer manufacturing companies; however, most small-scale manufacturers do not use software programs for transformer design due to the high cost of software. Instead, they rely on spreadsheets, such as Excel, for design calculations. This research aims to develop a free transformer design software program specifically tailored for small-scale manufacturers to eliminate trial and error in the design process. The paper presents four different transformer design methods and discusses their respective advantages and disadvantages. Firstly, the spreadsheet-based approach is converted into a software program, automating calculations for all possible cases. Secondly, the method involves generating all possible designs and selecting the best among them. The third approach utilizes parallel processing to enhance the efficiency of the second method. Lastly, a deep learning model is applied. The research findings demonstrate that the deep learning model, with inputs representing requirements like efficiency and outputs corresponding to necessary design parameters, operates with high accuracy. For future research, the plan is to expand the deep learning model to consider various input requirements, including weight, volume, price, voltage, loss, and wiring method. Moreover, the output layer representing design parameters will be extended to provide effective solutions for a wider range of design problems. These efforts are expected to innovate the transformer design process and contribute to energy conservation.