Topological optimal design of composite magnetic actuators to improve driving force and thermal conductivity

This study aims to introduce a topology optimization approach to enhance the driving force of magnetic actuators along with minimizing operating temperatures considering the nonlinearity of composite materials. The anisotropic magnetic composite comprises two distinct materials, considering differences in magnetic saturation effect and thermal conductivity. The first component exhibits low magnetic reluctivity and high thermal conductivity, while the other component displays high reluctivity and low conductivity. The representative volume element method (RVE) and deep neural network (DNN) were employed to obtain a dataset of effective composite material properties and to generate a machine learning (ML) module for determining material properties during the optimization process. To optimize and validate both performances, a multi-objective function was formulated. Utilizing an adaptive weighting method that gradually adjusts preferences from the initial to utopia point, the design process was performed to obtain Pareto-optimal solution sets faster while ensuring their even distribution in the objective space. Numerical examples are provided aimed at validating the proposed design process. The design results when applying high and low currents were compared to investigate nonlinear effects due to the magnetic saturation effect.