Optimization of Material and Operational Costs in Electromagnet Design

This study addresses the problem of optimizing the design of an electromagnet with a straight armature by considering both material and operational costs. A hybrid optimization framework was developed, combining traditional heuristic optimization techniques with machine learning-based models. Neural networks were trained on a dedicated database to predict the objective functions, namely material cost and power consumption, as functions of the design variables and the temperature of the control coil winding. The performance metrics for the predictive models demonstrate high accuracy (R² > 0.9). Both single-objective and multi-objective formulations were analyzed, with multi-objective optimization implemented through a weighted scalar function. The proposed method enables an efficient search for optimal solutions while reducing computational effort. The results confirm the potential of integrating data-driven models with optimization strategies in the design of cost-effective and energy-efficient electromagnetic systems. The proposed framework provides designers with a structured, data-driven approach for balancing manufacturing costs and operational efficiency.