.png){kind=icon}
.png){kind=icon}
Return
Electromagnetic Performance Analysis, Prediction, and Multiobjective Optimization for U-Type IPMSM
.png){kind=icon}
PRE
AI
DOI：10.1109/TIE.2023.3342326.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A novel multiobjective optimization model is presented for the interior permanent magnet synchronous motors (IPMSMs). First of all, in the model initialization stage, appropriate design variables are determined for the actual topology. Also, with reference to the engineering needs, the key electromagnetic characteristics relating to the no-load and on-load magnetic fields are selected as the main optimization objectives in order to achieve global performance improvement. Next, in the model prediction stage, two performance prediction models are proposed, studied, and implemented in parallel. One is an analytical model (AM) based on the improved subdomain approach and the magnetic equivalent circuit. It is utilized to predict the no-load electromagnetic characteristics of IPMSM. The complex structure and core nonlinearity of IPMSM are also reasonably accounted for. The other is a surrogate model (SM) based on the intelligent machine learning language (support vector regression). It is utilized to predict the on-load electromagnetic characteristics of IPMSM. The reliance on finite-element analysis is further also minimized. The proposed AM and SM have commendable behavior in terms of analysis speed, prediction accuracy, and storage consumption. Meanwhile, AM screens credible samples as well as provides robust support for the construction of SM. All of these signs build a solid foundation for a substantial boost in optimization efficiency. Afterward, in the model optimization stage, the advanced nondominated sorting genetic algorithm III is investigated to complete the final multiobjective optimization. Ultimately, a large number of calculations, simulations, and experiments have highlighted the effectiveness, rationality, and engineering practicality of this research.
Keywords:
Predictive models
Electromagnetics
Optimization methods
Synchronous motors
Magnetic circuits
Topology
Magnetic domains
Fast performance analysis and optimal design
improved subdomain approach
interior permanent magnet synchronous motor (IPMSM)
machine learning language (support vector regression)
nondominated sorting genetic algorithm III (NSGA-III)
Journal
IF:
7.2
Papers:
1.8W
Citations:
9.8W