AI Data-Driven Framework for Optimal Electrical Machine Design in Aerospace Application

Nowadays, a permanent magnet synchronous motor (PMSM) has emerged as a potential alternative for aerospace actuator applications. The PMSM is a high-power-to-weight-ratio electrical machine and can provide reliable operation within the required thermal limits. An electrical machine designer applies multiphysics optimization techniques to optimize the parameters of an electrical machine. The optimization process is combined with electromagnetic and thermal finite element analysis (FEA) to generate feasible machine designs. FEA is computationally expensive and time-consuming, which makes it difficult to analyze over varied geometrical parameters and increases the likelihood of selecting a suboptimal electrical machine design. This work introduces the application of AI in machine design & selection of optimal machine dimensions, which reduces the requirement of FEA for an increased number of variations in geometric parameters. Three AI models have been developed for this purpose. The first AI model has been developed to predict the output parameter based on nine types of geometric input parameters. The output parameters are treated as input for the second AI model to predict whether this configuration is feasible or non-feasible. The third AI data-driven model uses clustering of the output parameters of feasible machines to identify the most suitable machine according to user requirements. The application of AI achieved an improvement of 31% in torque density of PMSM. The selected optimal electrical machine has been prototyped and validated through experimental tests. Multiple design is also possible to obtain using the same framework. To showcase additional design through the framework, two additional machine design having different geometries for increase in torque & decrease in total weight and decrease in torque ripple, & magnet weight have been presented.