Digital design

This article provides an overview of how artificial intelligence (AI) is applied in designing high-frequency magnetic components, primarily high-frequency inductors and transformers, for power electronics systems. Four categories of AI, including expert systems, fuzzy logic, metaheuristic methods, and machine learning techniques, are addressed. First, AI models for estimating losses in high-frequency magnetic components are discussed. Subsequently, AI-based design methods in high-frequency inductors and transformers are observed. Then, AI tools applied to the automatic design of high-frequency magnetic components are introduced and compared. Drawing insights from an analysis of over 200 publications, this article highlights significant advancements: the development of AI-driven models for precise loss estimation in high-frequency magnetic components, the application of AI in optimizing design configurations for the components, and the automation of design processes. These achievements demonstrate AI's capability to enhance the efficiency, performance, and innovation in high-frequency magnetic component design, offering a roadmap for future research in power electronics systems.

This article tackles the complex challenge of predicting magnetic iron losses in high-frequency magnetic components by introducing a novel conditional generative adversarial network model. Diverging from traditional loss prediction methodologies that often overlook intricate interactions of factors, our conditional generative adversarial network framework is designed to comprehensively incorporate diverse aspects such as material properties, geometrical variations, and environmental conditions. To facilitate this advanced approach, a specialized four-wire measurement kit was employed, which significantly enriched the training dataset with a wide range of measurements. When benchmarked against conventional deep neural network models, the conditional generative adversarial network not only achieves faster convergence but also demonstrates markedly superior accuracy in predicting iron losses. This superiority is particularly notable in scenarios that extend beyond the training data's range, underscoring the model's robustness and adaptability. Such advancements in predictive accuracy and efficiency represent a significant leap forward in the design and optimization of high-frequency magnetic components.

This paper presents a non-invasive, real-time fault detection and predictive maintenance framework for Connected Electric Truck powertrain. Leveraging the ultra-low-power ISM330DHCX sensor’s Machine Learning Core (MLC), the system performs on-sensor processing of accelerometer and gyroscope signals using Edge AI and TinyML techniques. Decision Trees (DTs) and an eight-DT based Random Forest (RF) model were employed to classify vibration patterns under varying operating conditions. Five experiments were conducted on electric motor test benches to simulate rotor imbalance and gather data in both normal and faulty states. Vibration signals were first classified into three broader fault categories and then mapped into 12 subclasses using a novel output mapping scheme. Guided by the Design, Implementation, Potential Failure and Functional Failure (DIPF) curve, this framework enables accurate detection of rotor unbalance faults. Experimental results showed that while single-DT models achieved high training accuracies (95−98.42%) but poor validation performance (<50%), the eight-DT based Random Forest with subclass splitting significantly improved generalization, achieving 98.74% validation accuracy. By minimizing power consumption and bandwidth requirements, this on-sensor Edge AI sensor network approach offers a scalable solution for predictive maintenance, with the potential to reduce downtime and maintenance costs by 30−50%. The findings highlight the promise of ultra-low-power intelligent sensing for enhancing the reliability and efficiency of Connected Electric Trucks.

This report presents a comprehensive overview of the activities carried out in Work Package 4 (WP4), specifically Task 4.3, of the RHODAS project, which focuses on the design and development of a digital twin framework for key components of electric powertrains.  The developed framework comprises three individual digital twins: the inverter digital twin, the e-motor digital twin, and the gearbox digital twin. The inverter digital twin models the electrothermal behavior of power electronics modules (SiC and GaN) to support performance monitoring and failure prediction. The e-motor digital twin captures the thermal behavior of the rotor and stator of the RHODAS e-motor, enabling accurate simulation under various driving conditions. The gearbox digital twin is based on a data-driven thermal equivalent circuit model that estimates the gearbox oil temperature using a range of input data.  All digital twin models are implemented in both MATLAB/Simulink and C++ to enable seamless integration into the RHODAS cloud platform, supporting both online and offline thermal monitoring and efficient computation.  The deliverable also includes extensive sensitivity analyses to investigate the behavior of the e-motor, power inverter, and gearbox under different operating modes. This outcome delivers a robust and scalable digital twin architecture that enhances monitoring, diagnostics, and predictive maintenance capabilities within the RHODAS electric vehicle platform. 

Developments of the digital twinning framework for electric drive design in the EU-funded project EM-TECH. 

Details about the distributed XIL testing methodology out of the research of the EU-funded project EM-TECH. 

Modern Smart Grids are complex systems incorporating physical components like distributed energy resources and storage, along with cyber components for advanced control, networking, and monitoring. This study proposes an integrated methodology for risk prioritization and failure mode classification into low, moderate and high-risk faults using Grey Relational Analysis (GRA) together with Failure Mode and Effects Analysis (FMEA) and Deep Learning algorithms. The results demonstrate that, especially in complex systems like Smart Microgrids, the proposed method more accurately captures the coupling relationships between failure modes compared to the conventional FMEA method.

31% of failures in power electronics are caused by semiconductor switch failure. Semiconductor junction temperature is the most crucial failure stress factor. A Remaining Useful Lifetime (RUL) estimation platform is needed for Prognostic and Health Management of power electronics converters.

The RHODaS Webinar Series presents four interconnected sessions exploring the latest European research on next-generation electric powertrain technologies. Hosted by the RHODaS consortium under the Horizon Europe framework, funded by ‪‪the European Commission‬ and as part of the E-VOLVE Cluster, the webinar series will feature insights from the RHODaS, SCAPE, ‪‪Maxima‬ and EM-TECH projects, bringing together leading experts in power electronics, digital systems, and sustainability. Each webinar focuses on a specific technological domain critical to the electrification of transport — from component design and thermal management to digital intelligence and circularity. Together, they illustrate how European research is transforming electric mobility through efficiency, reliability, and environmental responsibility. 

Digitalization is redefining how electric vehicles are monitored and optimized. Experts from RHODaS, SCAPE, and MAXIMA will showcase digital twin technologies, fault management systems, and real-time health assessment tools for next-generation powertrains.

The study introduces a proof-of-concept self-adaptive neural network model predictive control (SA-NNMPC) system, which uses a neural network as main component of the prediction model, for the anti-jerk control of electric vehicles. Through the adaptation mechanism of the network and cost function weights during vehicle operation, which is activated when the plant behaves significantly differently from its digital twin, the SA-NNMPC architecture adjusts to the progressive vehicle aging, or to the replacement of hardware parts, which is expected to be an important feature of next-generation vehicles. Validation tests and simulations show that the neural network accurately replicates the drivetrain dynamics of the considered electric vehicle, and, for nominal conditions, already leads to a performance improvement of the NNMPC implementation – which can run in real-time on a rapid control prototyping unit – with respect to a benchmarking nonlinear model predictive anti-jerk controller. Moreover, the preliminary simulation results confirm the potential of the proposed architecture in terms of: i) adaptability to operating conditions not covered in the original training, and variations of vehicle parameters; and ii) auto-tuning of the algorithm when applied to different vehicles.