Design parameter optimization of Electric Drive Units. A regression based optimization of PMSM geometrical design parameters and final drive ratio
Typ
Examensarbete för masterexamen
Master's Thesis
Master's Thesis
Program
Mobility engineering (MPMOB), MSc
Publicerad
2023
Författare
Panyam, Adithya Ram
Panguluru, Chanakya
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Permanent magnet synchronous machines (PMSM) are widely used for propulsion system in electric vehicles due to their high efficiency, high torque density, excellent dynamic response and control, and reliability. Significant investments are being made to fund research on improving their efficiency, torque capabilities to enhance overall vehicle performance and driving range. To appropriately determine the size and specifications of the motor for a specific powertrain requirement, numerous geometric, electric, and mechanical parameters need to be considered. In this study, we present a method to optimize certain selected design parameters of the PMSM
to improve the performance of the powertrain. The focus of this study is on the geometrical design parameters of the PMSM, while also considering the final drive ratio, to imrprove the performance of the overall powertrain. By employing orthogonal design of experiments, simulations are conducted with predefined levels of these parameters to generate a dataset used for training regression models. Two types of regression models, linear and Gaussian process, are considered and compared. The study reveals that Gaussian process regression provides more accurate predictions for the selected output variables. Subsequently, the Gaussian process regression model
is used for optimizing the design parameters. The optimization process incorporates cost functions defined for particular application, such as efficiency optimization for Passenger variant and performance optimization for Performance variant.
Beskrivning
Ämne/nyckelord
permanent magnet , synchronous machine , parameter study , Gaussian process regression , linear regression , sensitivity analysis , design optimization