Deep Learning-Based Prediction of Key Performance Indicators for Electrical Machines

Abstract

The design of an electrical machine can be quantified and evaluated by Key Performance Indicators (KPIs) such as maximum torque, critical field strength, costs of active parts, sound power, etc. Generally, cross-domain tool-chains are used to optimize all the KPIs from different domains (multiobjective optimization) by varying the given input parameters in the largest possible design space. This optimization process involves magneto-static finite element simulation to obtain these decisive KPIs. It makes the whole process a vehemently time-consuming computational task that counts on the availability of resources with the involvement of high computational cost. In this paper, a data-aided, deep learningbased meta-model is employed to predict the KPIs of an electrical machine quickly and with high accuracy to accelerate the full optimization process and reduce its computational costs. The focus is on analyzing various forms of input data that serve as a geometry representation of the machine. Namely, these are the cross-section image of the electrical machine that allows a very general description of the geometry relating to different topologies and the the classical way of scalar geometry parametrizations. The impact of the resolution of the image is studied in detail. The results show a high prediction accuracy and proof that deep learning-based meta-models are able to minimize the optimization time. The results also indicate that the prediction quality of an image-based approach can be made comparable to the classical way based on scalar parameters.

Optimization of an IPMSM for Constant-Angle Square-Wave Control of a BLDC Drive

“…The presented optimization in this study was also performed based on a metamodel trained with a feedforward neural network. Exchanging a computationally expensive numerical evaluation with a neural network was presented in [29][30][31]. A neural network metamodel was used in a multi-objective optimization in [32].…”

Learning-Based Approaches to Current Identification from Magnetic Sensors

“…Recently, many research areas have taken advantage of the potential offered by behavioral models based on machine learning (ML) or (deep) neural networks (DNN) [1,2]. In fact, recent works on the DNN-assisted analysis of electromagnetic (EM) field computation problems showed the promising potential of convolutional neural networks (CNN) and ML tools [3][4][5][6][7][8][9][10][11]. A comprehensive review of recent works on ML for the design optimization of electromagnetic devices can be found in [4], where the growing interest of the community is clearly evidenced.…”

“…A comprehensive review of recent works on ML for the design optimization of electromagnetic devices can be found in [4], where the growing interest of the community is clearly evidenced. Some works adopted ML or DNN models to predict the key performance indicators of electrical machines [5][6][7], whilst others focused on topology optimization [8][9][10][11].…”

A Regularized Procedure to Generate a Deep Learning Model for Topology Optimization of Electromagnetic Devices

“…Recently, many research areas have benefited from the potential offered by deep learning (DL) tools, such as convolutional neural networks (CNNs) [1,2]. In fact, recent works on the DL-assisted analysis of electromagnetic (EM) field computation problems showed the promising potential of CNN applications [3][4][5][6][7][8][9][10][11][12][13][14][15].…”

“…A comprehensive review of recent works on machine learning for the design optimization of electromagnetic devices can be found in [4], where the growing interest of the community for DL is clearly evidenced. Some works have adopted DL models to predict the key performance indicators of electrical machines [6,8,9,15], whilst others have focused on topology optimization [10][11][12][13].…”