This paper provides an overview of current Noise Vibration and Harshness (NVH) challenges for the design of Permanent Magnet Synchronous Motors (PMSMs) for automotive applications. Particular attention is devoted to analytical and numerical methods utilized in NVH modelling. Electromagnetic force computation, motor structural modelling and acoustic emission calculation are reviewed focusing on NVH emissions. The literature reveals, as expected, that analytical methods are simpler than numerical ones, and are usually employed during the preliminary design stage. Numerical methods are adopted in the detailed design stage as the computational cost increases. PMSMs are selected for high performance and high efficiency motor drives, they are widely adopted in the actual transition to vehicle electrification. The paper provides guidelines for facing NVH design issues of PMSMs.

In this paper, harmonic components are injected into the phase current of a Permanent Magnet Synchronous Motor (PMSM) used in automotive applications to improve the Noise Vibration and Harshness (NVH) performance. The study proposes and validates a numerical method for modeling this phenomenon. The strategy was first applied experimentally. Subsequently, the experimental currents were measured and used as input for the numerical models developed in Flux2D® and Abaqus®. The harmonic contributions of the experimental currents, obtained from the inverter control strategy, are analysed and adopted as inputs to the electromagnetic model to calculate the forces. These are applied to the Finite Element Model (FEM), which returns the values of vibration and acoustic emission of the motor as the injection strategy is varied. The numerical results are then compared and validated with the experimental data. The strategy is applied as an NVH countermeasure to a specific time order that is particularly noisy for the analyzed PMSM. This order turns out to be dominated by a spatial 0 mode also referred to as the breathing mode. The phase of the injected current harmonic is varied between 0 and 2π. For a specific phase angle, a minimum of the acoustic emission level is reached. The application of the strategy experimentally shows a maximum reduction of 29 dB and 27 dB from the vibration and acoustic levels respectively. Numerical models are able to capture the physics of the system, reproducing the variation in acoustic emission as a function of the phase angle of the injected current. Numerical results are close to the experimental data, with a maximum difference of 3dB on the optimal vibration reduction condition.

Fault detection is a critical research area, especially in the automotive sector, aiming to quickly assess component conditions. Machine Learning techniques, powered by Artificial Intelligence, now represent state-of-the-art methods for this purpose. This study focuses on durability testing of Permanent Magnet Synchronous Motors for automotive applications, using Autoencoders (AEs) to predict and prevent failures. This AI-based fault detection strategy employs acceleration signals coming from electric motors tested under challenging conditions with significant variations in torque and speed. This approach goes beyond typical fault detection in steady-state conditions. Based on a review of Neural Networks, including Variational Autoencoders (VAEs), Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, the performance of six AI architectures is compared: AE, VAE, 1D CNN AE, 1D CNN VAE, LSTM AE and LSTM VAE. The 1D CNN AE outperformed the other networks in fault detection, showing high accuracy, stability and computational efficiency. The model is integrated into an algorithm for semi-real-time fault monitoring. The algorithm effectively detects potential motor failures in real-world scenarios, including bearing faults, mechanical misalignments, and progressive wear of components, thereby proactively preventing damage and halving test bench downtime.

This paper deals with an adaptive multi-objective optimisation process for the design of electric motors by means of supervised learning techniques. The process is based on a multi-objective optimisation approach which involves many objective functions (mass, rotor inertia, average total losses), constraints (geometrical feasibility, current and voltage limit, torque-speed profile, thermal constraints) and a large number of design variables (rotor, stator and winding parameters) that describe the physical properties of the motor. The design of computer experiments is based on a Low Discrepancy Sequence (LDS). The electric motor performance indices are evaluated through multiphysics simulations (electromagnetic, thermal) carried out by Motor-CAD™ software. Artificial Intelligence (AI) is adopted to approximate the physical model behaviour. Artificial Neural Networks (ANN) are exploited, in this way a large set of design variables combinations can be investigated with a reasonable computational effort. Since the multi-objective optimisation of electric motor is particularly complex, a special algorithm had to be developed to find Pareto-Optimal solutions. A new Adaptive Pareto Algorithm allows to reduce the computational cost and to achieve an even distribution of the optimal solutions in the Pareto optimal front. The algorithm is particularly effective when it is integrated with a global approximation model, the infill operation allows to improve the metamodel approximation. The derived electric motors show the best compromise performances among millions of possible configurations.

The automotive industry is experiencing a period of transition from traditional internal combustion engine (ICE) vehicles to electric vehicles. Although electric machines have always been used in many applications, they are generally designed neglecting the sources of uncertainty, even such uncertainty can lead to significant deterioration of the motor performance. The aim of this paper is to compare the results obtained from the multi-objective optimization of an interior permanent magnet synchronous motor (IPMSM) using a robust approach versus a deterministic one. Unlike other studies in the literature, this research simultaneously considers different sources of uncertainty, such as geometric parameters, magnet properties, and operating temperature, to assess the variability of electric motor performance. Different designs of a 48 slot–8 pole motor are simulated with finite element analysis, then the outputs are used to train artificial neural networks that are employed to find the optimal design with different approaches. The method incorporates an innovative use of the neural network-based variance estimation (NNVE) technique to efficiently calculate the standard deviation of the objective functions. Finally, the results of the robust optimization are compared with those of the deterministic optimization. Due to the small margin of improvement in robustness, both methods lead to similar results.

With the growing demand for hybrid and electric vehicles, the accurate prediction of NVH (Noise, Vibration, and Harshness) behavior in Permanent Magnet Synchronous Machines (PMSMs) has become a critical aspect of electric motor design. This paper presents a detailed modeling approach for electromagnetic-induced noise and vibrations in PMSMs, integrating both analytical and numerical methods. The model focuses on quantifying the contributions of radial and tangential electromagnetic forces, which are key drivers of vibro-acoustic responses. The analytical part employs curved beam theory and a simplified acoustic model, offering rapid insights during early design stages. In parallel, a detailed numerical model based on finite element analysis is developed using a physics-based approach that accounts for the actual geometry and material properties of the PMSM prototype. This allows for enhanced accuracy without relying on experimental material parameter identification. Moreover, the detailed model includes the fluid–structure interaction introduced by the channels of the cooling fluid of the electric machine, which, although poorly addressed by the existing literature, was found to play a key role in driving the vibrational behaviour of the structure. By combining analytical speed with numerical precision, the proposed approach enables consistent and physically-based NVH predictions across various design phases, ultimately supporting improved electric machine performance and reducing development time and costs. Validation against experimental data confirms the ability of the model to accurately predict both sound pressure levels and housing surface vibrations. The novelty of this work lies in its integration of fluid–structure interaction and material modeling without the need for empirical parameter tuning, offering a robust tool for NVH design in electric vehicle applications.