Reduced-Order Eddy-Current Loss Modelling of Electrical Machines Using Lookup Tables Obtained via 2D Finite-Element

Pinto, Diogo E.; Pop, Adrian-Cornel; Kempkes, Joachim; Gyselinck, Johan

doi:10.1109/icelmach.2018.8507025

Cited by 6 publications

(4 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Exploiting periodicity, it suffices to cover one third of an electrical period; this is done with 60 discrete rotor positions. Note that such a magnetostatic characterization also allows for a-posteriori iron losses estimation, and that magnetodynamic (time-stepping) FE analysis is less relevant for PMSMs [19].…”

Section: B Training Data Generationmentioning

confidence: 99%

Machine-Learning Aided Multiobjective Optimization of Electric Machines— Geometric-Feasibility and Enhanced Regression Models

Pop

Cai²,

Gyselinck

2023

IEEE J. Emerg. Sel. Top. Ind. Electron.

Self Cite

View full text Add to dashboard Cite

This paper deals with the optimization of electrical machines by means of ANN-based classification and regression models. Geometrically (or otherwise) unfeasible designs are detected with high accuracy during the optimization process by means of an ad hoc classification model, whereas continuous targets are predicted through regression models. Training samples are generated with an expensive finite-element (FE) model, resulting in small training sets. Moreover, the design optimization normally involves multiple but correlated sub-objectives. The correlation can be leveraged using chained regression or a multioutput ANN; it is shown that both methods can achieve higher predictive performance than predicting the targets separately. The developed methods are successfully applied to a permanentmagnet synchronous machine (PMSM) with 12 geometric parameters and 4 sub-objectives, and considering both no-load and load operation. The results show very good predictive performance of the ANN models and a significant reduction of the computational effort. The optimization can thus be run several times, with e.g. modified weighting of the sub-objectives, with little extra cost.

show abstract

Section: B Training Data Generationmentioning

confidence: 99%

Machine-Learning Aided Multiobjective Optimization of Electric Machines— Geometric-Feasibility and Enhanced Regression Models

Pop

Cai²,

Gyselinck

2023

IEEE J. Emerg. Sel. Top. Ind. Electron.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The motor parameters (i.e., quadrature inductances) are obtained as a function of d-and q-axis currents from FEA using frozen permeability method (FPM) [11]- [12]. The inductances are included in the dynamic model using 2D LUTs.…”

Section: Flux-linkage Vs Linear Ann-based Modelsmentioning

confidence: 99%

Artificial Neural Network and Data Dimensionality Reduction Based on Machine Learning Methods for PMSM Model Order Reduction

et al. 2021

View full text Add to dashboard Cite

The present paper targets a solution for permanent motor synchronous machine (PMSM) model order reduction (MOR) using artificial neural networks and machine learning techniques for data dimensionality reduction. The neural networks are trained using data obtained from a series of electromagnetic Finite Element Analysis (FEA), conducted in conditions imposed by the data dimensionality reduction method. The workflow proposed to build the PMSM MOR, starts with data generation, goes further to its post-processing, and finishes with the model training and experimental validation. In the study, data dimensionality reduction procedure (adaptive data generation) is performed to increase the computational efficiency, also maintaining the model accuracy. Different data reduction approaches are compared from the computational cost's point of view and their ease of use. The obtained results are compared to those obtained from FEA seeking the best solution for building the dynamic model. The resulting ROM is included in a real-time control prototyping platform to characterize machine's performances. The model accuracy and its usability are proved in a comparative analysis with simulated versus experimental measurements.INDEX TERMS artificial neural networks, adaptive data generation, machine learning, permanent magnet synchronous machine, dynamic simulation, real-time.

show abstract

“…Section III explores two types of CSVMs based on magnetic co-energy [3], [4] and based on flux-linkages [5], [6] and electromagnetic torque [7] data obtained from FE. Section IV explores two types of FSVMs [8], where the difference resides in the fluxlinkage map inversion technique. For an accurate time-domain NVH simulation, two main attributes are important in the voltage-fed PMSM model: the electromagnetic torque because it represents one of the main sources of the structure-born noise together with the unbalanced magnetic pull (UMP) [9] and the computed currents that represent inputs to the force models, representing the main source of the air-born noise [1].…”

Section: Introductionmentioning

confidence: 99%

A comparative study of system-level PMSM models with either current or flux-linkage state variables used for vibro-acoustic computation

Ciceo

Chauvicourt

Gyselinck

et al. 2019

2019 IEEE International Electric Machines &Amp; Drives Conference (IEMDC)

Self Cite

View full text Add to dashboard Cite

This paper explores different high-fidelity, systemlevel models for permanent magnet synchronous machines. The main attributes examined are electromagnetic torque and forces representing the main contribution sources for noise and vibration in electrical powertrains. First, two types of electromagnetic models extracted from finite element analysis are presented: the current and flux-linkage state variable models. Afterwards, two force transformation models used for system-level vibroacoustic computation are explored and a spectral convergence test is proposed. Procedures regarding model implementation, such as look-up table derivation/inversion are discussed and design trade-offs are analyzed. Finally, the different model results are benchmarked against a finite element model and the real-time performance is presented.

show abstract

Reduced-Order Eddy-Current Loss Modelling of Electrical Machines Using Lookup Tables Obtained via 2D Finite-Element

Cited by 6 publications

References 14 publications

Machine-Learning Aided Multiobjective Optimization of Electric Machines— Geometric-Feasibility and Enhanced Regression Models

Machine-Learning Aided Multiobjective Optimization of Electric Machines— Geometric-Feasibility and Enhanced Regression Models

Artificial Neural Network and Data Dimensionality Reduction Based on Machine Learning Methods for PMSM Model Order Reduction

A comparative study of system-level PMSM models with either current or flux-linkage state variables used for vibro-acoustic computation

Contact Info

Product

Resources

About