摘要: Electrical motors play a crucial role in many industrial applications, making them asignificant research area.Maximizing their potential is the focus of many studies and is usually achieved throughgeometry optimization and materia... 展开 Electrical motors play a crucial role in many industrial applications, making them asignificant research area.Maximizing their potential is the focus of many studies and is usually achieved throughgeometry optimization and material selection. However, pulse width modulation (PWM)command optimization, while sometimes overlooked, is also an efficient approach toincrease motor efficiency and reduce torque ripple or noise emissions.Numerical simulation considering non-linearities or harmonic phenomena is a key toolto support this optimization. Nevertheless, full-scale electrical motor models are toocomputationally expensive. To overcome these limitations, efficient model reductiontechniques preserving accuracy are needed.In this study, we present a novel approach for reducing electrical motor models thatcombines both numerical simulation and analytical methods. By means of finite element(FE) models, relevant electromagnetic quantities such as flux density, flux linkages,torque, and forces are calculated and used to calibrate the reduced motormodels.Load profiles have been simulated with both FE coupled with an electric circuit and areduced time-based model. Results have shown high fidelity and a reduction of thecomputation time by a factor of 1,000.In addition, to further reduce the computation time by a factor of 10, a frequency-basedmodel has been developed. The model reduction is based on local parameter linearizationbut still preserving the information about machine saturation, cross-saturation,and spatial harmonics.Moreover, a method for a fast calculation of motor losses have been developed. Thismethod can be coupled to both time-based and frequency-based reduced models.All the above works have been evaluated against experimental results on a full-scale150 kW powertrain showing the interest of the approach.In the end, the above-mentioned models make a very powerful tool supporting ripplereduction strategies as well as system efficiency optimization.Those works were carried out within the OLEA-EVO project,which is conducted in collaboration with Silicon Mobility. This project was financed bythe French government within the framework of the Investments for the Future Program. 收起