Lin Liu

Despite the urgent demand in industry, there do not yet exist practical and accurate simulation methods for calculating the magnetic losses of ferromagnetic laminated cores in electrical machine (EM) simulations. Transfer learning, a machine learning technique, utilizes a model developed for one task as a starting point for developing a model for another task. This study innovatively integrates parametric homogenized material laws with deep learning (DL) approaches. Firstly, the intrinsic resistance parameters characterizing core material properties are investigated through the combined application of the Bertotti model and an improved equivalent circuit model. Experimental data from prototypes are then used to visualize the influence patterns of different temperatures, harmonics, loads, and frequencies on these derived parameters. To both reduce the required data for improved performance and model the underlying mechanisms of mutual parameter impact, a sizeable auxiliary dataset was synthetically generated through the synergistic application of support vector machine, decision tree, and backpropagation (BP) neural network algorithms. By leveraging the pre-learned patterns of parameter variation, auxiliary data, and a small set of target data covering high-frequency, high-temperature, and high-load conditions, a two-stage feature transfer learning process is implemented using deep sparse autoencoder and BP neural network. Finally, comparative experiments are designed to verify the limitations of existing loss evaluation methods and to emphasize the importance of comprehensively considering multi-factor coupling effects in EM core loss modeling. The results demonstrate that the proposed local material law can serve as a conventional constitutive relationship in two-dimensional or three-dimensional modeling, and, when combined with DL algorithms, achieves a massive speed-up compared to monolithic coupling.

The precise calculation of iron losses in permanent magnet synchronous motors (PMSMs) remains challenging due to the interplay between various disciplines such as electromagnetism, magnetism, and thermal/mechanical dynamics. Purely mechanistic models require detailed theoretical knowledge and exact parameters, often struggling to accurately describe complex systems, while purely data‐driven methods lack interpretability, which are susceptible to data noise and outliers in feature extraction and complicated pattern recognition. Consequently, this paper aims to present a hybrid mechanism‐data‐driven model for accurately estimating the iron loss for PMSMs, considering the multiphysics coupling effects. Specifically, based on the well‐defined physical principles, an advanced iron loss analytical model that simultaneously considers mechanical stress, temperature rise, harmonics, load currents, and changing frequency is developed and then utilised to calculate numerous loss data under different operating conditions, providing a certain level of stability and reliability for prediction accuracy. Subsequently, a convolutional neural network (CNN) algorithm is employed to perform deep learning to extract features and patterns from the data. By defining a suitable loss function, the pre‐trained model was fine‐tuned and optimised using a small amount of actual data. To validate its superiority, extensive numerical and experimental analyses are conducted on the prototype. The results demonstrate that the iron losses computed using this hybrid model overcome the limitations of singular methods by effectively leveraging both theoretical knowledge and real‐world data, thus accurately accommodating various application scenarios. This integrated approach enhances the accuracy, stability, and interpretability of the model, laying a solid foundation for more specialised applications in the future.

Electrical machine optimisation is normally a high‐dimensional non‐linear multi‐objective optimisation problem. A multi‐level optimisation (MO) strategy is currently used to improve efficiency, where sensitivity analysis is required for dividing design parameters into different groups. However, the conventional MO strategy cannot handle ultra‐high‐dimensional optimisation problems. In this paper, a sensitivity analysis method with variable weighted intervals is proposed to calculate the sensitivity coefficient in the parameter design range. Moreover, three improved multi‐level optimisation strategies based on different optimisation algorithms, sequential sensitivity strategies, and machine learning models are proposed, analysed, and compared with the conventional MO strategy. Through a case study of a synchronous reluctance machine, it can be seen that the proposed optimisation strategies can improve the optimisation results and efficiency of ultra‐high‐dimensional optimisation of electrical machines.

The loss performance of interior permanent magnet synchronous motors (IPMSMs), commonly employed in electric drive systems of electric vehicles (EVs), is notably influenced by temperature fluctuations and harmonics in the supply current, because the magnetic properties of silicon steel sheets are easily vulnerable to these factors. Therefore, in this paper, an advanced experimental setup is utilized to quantify the AC loss of silicon steel samples under rotating magnetic fields. Then, an enhanced analytical estimation model of the iron loss in IPMSMs is formulated by incorporating experimentally-fitted coefficients for hysteresis and eddy current losses, allowing for a comprehensive consideration of temperature variations and harmonics. Given that several calculation methods demonstrated satisfactory accuracy, when relying solely on the no-load iron loss, the efficacy of the proposed model is validated through loaded experiments. Moreover, comparative results verify the improved accuracy and computational simplicity of the presented model, even under the dynamic drive cycle conditions of EVs. The proposed tester and approaches can also be applied to the design optimization of electric motors with high-temperature superconducting (HTS) materials.

During the lifecycle of an electric drive system, the electricity cost usually constitutes a dominant portion, leading to extensive research on improving the efficiency of such systems. This article presents an efficient design optimization strategy for permanent magnet synchronous motor (PMSM) drive systems, cooperating with an improved loss minimization control, targeting to reduce the system power loss and improve the overall operating performances. Specifically, a generalized equivalent circuit model with a novel topology that considers rotating magnetization effects is first proposed to accurately predict the core loss of PMSMs under both no-load and load conditions. Moreover, after analyzing the copper and inverter losses, an integrated system-level loss model of the PMSM drive system is developed, based on which an improved loss minimization control is implemented through deriving the relationships among the electromagnetic torque, d-axis and q-axis currents. Then, the system-level optimization strategy that incorporates both the motor and control sides is presented, aiming to maximize the torque and efficiency and minimize the torque ripple and speed overshoot. Finally, a multilevel algorithm based on sensitivity analysis is employed to solve the multiobjective optimization models with reduced calculation costs. Extensive experimental and simulation verifications are demonstrated with a 20 kW interior PMSM drive system platform. Results show that the optimized drive system has promoted steady-state and dynamic behaviors, and the proposed loss minimization control can comprehensively reduce the system losses by about 3.25%.

With the increasing demand for smaller, lighter, and more affordable electromagnetic devices, there is a growing trend toward developing high-power-density transformers and electrical machines. While increasing the operating frequency is a straightforward approach to achieving high power density, it can lead to significant power loss within a limited volume, resulting in excessive temperature rise and device degradation. Therefore, it is crucial to design high-power-density electromagnetic devices that exhibit low power loss and efficient thermal dissipation to address these challenges. Advanced techniques, such as the utilization of novel and advanced electromagnetic materials, hold great promise for overcoming these issues. Specifically, nanocrystalline and amorphous magnetic materials have emerged as highly effective solutions for reducing power loss and increasing efficiency in electromagnetic devices. This paper aims to provide an overview of the application of nanocrystalline and amorphous magnetic materials in transformers and electrical machines, along with key technologies and the major challenges involved.

As we face issues of fossil fuel depletion and environmental pollution, it is becoming increasingly important to transition towards clean renewable energies and electric vehicles (EVs). However, designing electric motors with high power density for EVs can be challenging due to space and weight constraints, as well as issues related to power loss and temperature rise. In order to overcome these challenges, a significant amount of research has been conducted on designing high-power-density electric motors with advanced materials, improved physical and mathematical modeling of materials and the motor system, and system-level multidisciplinary optimization of the entire drive system. These technologies aim to achieve high reliability and optimal performance at the system level. This paper provides an overview of the key technologies for designing high-power-density electric motors for EVs with high reliability and system-level optimal performance, with the focus on advanced magnetic materials and the proper modeling of core losses under two-dimensional or three-dimensional vectorial magnetizations. This paper will also discuss the major challenges associated with designing these motors and the possible future research directions in the field.

The hybrid drive wind turbine (WT) can be friendly connected to the power grid by using a speed regulating differential mechanism (SRDM) instead of partially or fully rated converters, which has been considered as a promising solution for the stable consumption of large-scale wind power generation. To further improve the on-grid performance of hybrid drive WTs, this paper develops a multi-source power generation scheme, in which a hydrogen storage system (HSS) is integrated for mitigating the wind power generation intermittencies. The overall architecture and kinematic principles of the proposed wind–hydrogen generation system, called SRDM-based WT with HSS, are first analyzed. Then, the graphical descriptions of mathematical models are finalized via the Energetic Macroscopic Representation method, by which the physical characteristics and energy flow relationships are revealed. To ensure the economical and stable operation of the proposed wind–hydrogen scheme, an effective optimal allocation framework, considering the uncertainties from wind power output and load demand, is presented to HSS, targeting the maximum annual revenue. The effects of several key HSS parameters on the capacity allocation results are also investigated. Moreover, aiming at the different system working modes, an energy management approach is synthesized to achieve the interaction analysis and power supervision between energy sources and storage elements. Finally, experimental and simulation case studies are demonstrated. Results illustrate the effectiveness of the proposed approaches and the optimal performance for uninterrupted on-grid operation of the proposed wind–hydrogen energy system.

Electric motors play a crucial role in modern industrial and domestic applications. With the trend of more and more electric drives, such as electric vehicles (EVs), the requirements for electric motors become higher and higher, e.g., high power density with good thermal dissipation and high reliability in harsh environments. Many efforts have been made to develop high performance electric motors, such as the application of advanced novel electromagnetic materials, modern control algorithms, advanced mathematical modeling, numerical computation, and artificial intelligence based optimization design techniques. Among many advanced magnetic materials, soft magnetic composite (SMC) appears very promising for developing novel electric motors, thanks to its many unique properties, such as magnetic and thermal isotropies, very low eddy current loss, and the prospect of low-cost mass production. This paper aims to present a comprehensive review about the application of SMC for developing various electric motors for electric drives, with emphasis on those with three-dimensional (3D) magnetic flux paths. The major techniques developed for designing the 3D flux SMC motors are also summarized, such as vectorial magnetic property characterization and system-level multi-discipline robust design optimization. Major challenges and possible future work in this area are also discussed.

The magnetic properties of magnetic cores are essential for the design of electrical machines, and consequently appropriate mathematical modeling is needed. Usually, the design and analysis of electrical machines consider only the one-dimensional (1D) magnetic properties of core materials, i.e., the relationship of magnetic flux density (B) versus magnetic field strength (H), and their associated power loss under 1D magnetization, in which the B and H are constrained in the same orientation. Some studies have also been performed with the two-dimensional (2D) magnetizations in which the B and H are vectorial, rotating on a plane, and they may not be in the same direction. It has been discovered that the 2D rotational property is very different from its 1D alternating counterpart. However, the magnetic fields in an electrical machine, in particular claw pole and transverse flux machines, are naturally three-dimensional (3D), and the B and H vectors are rotational and may not lie on the same plane. It can be expected that the 3D vectorial property might be different from its 2D or 1D counterpart, and hence it should be investigated for the interests of both academic research and engineering application. This paper targets at a general summary about the magnetic material characterization with 3D vectorial magnetization, and their application prospect in electrical machine design and analysis.