Real-time Virtual Prototypes for the Power Electronics Supply Chain

电力电子供应链的实时虚拟原型

• 批准号: EP/X024377/1
• 负责人: Paul Evans
• 金额: $ 132.08万
• 依托单位: University of Nottingham
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FX024377%2F1
• 关键词: Real; time; Virtual; Prototypes; Power

This project investigates computer simulation methods for power electronic systems. Power electronic systems are essential sub-systems in key energy conversion application areas such as electric vehicle powertrains, marine propulsion, aerospace, renewable energy and power distribution. They are complex assemblies of electrical, mechanical and thermal management sub-systems and components. Optimal system designs require understanding of electrical, electromagnetic and thermal interactions between components - the way in which a component is integrated during system manufacture can have a significant effect on system performance and lifetime. Computer models that can be passed from component to system manufacturers are needed to allow effective digital system design optimisation. Existing models provided by power electronic component manufacturers are limited to circuit models which cannot account for the 3D system geometry, component placement, or manufacturing processes used. 3D CAD component models could be provided but to be useful, detailed and high-resolution models are needed which would expose IP. Complex Finite Element simulations would then be needed to evaluate these models and these simulations are extremely computationally expensive - potentially taking days to complete. Historically, models have been used to evaluate worst case electrical and thermal performance given expected operating conditions but increasingly, lifetime and reliability is of concern. Predicting worst-case electrical and thermal performance is straightforward because maximum power and ambient temperature operating points can easily be defined and simulated. Predicting lifetime and service intervals for components is more difficult as component wear-out is determined by accumulated stress and damage sustained under normal operating conditions - different conditions within the acceptable performance envelope can give drastically different service lifetimes. Wear-out also occurs over long time periods which necessitates long simulations, if the models used are not incredibly efficient then this further increases the amount of time required to run the simulations.The research undertaken will propose a new Real-Time Virtual Prototype (RTVP) model architecture for power electronic components. The RTVP models utilise reduced order modelling algorithms that allow the models to simulate over 1000 times faster than conventional Finite Element models. These models can then be coupled together and simulated very quickly (faster than real-time in certain scenarios) to allow system manufacturers to evaluate system performance, including wear-out and reliability, over extended time periods. Furthermore, the models can be configured to hide sensitive design and performance data which will enable component manufacturers to release accurate, 3D models simulation models of their components whilst protecting sensitive IP. These models can be combined to produce full digital "virtual prototypes" of system designs, eliminating the need for construction and testing of physical prototypes, leading to reduced design costs and increased system performance.

本课题研究电力电子系统的计算机仿真方法。电力电子系统是电动汽车动力系统、船舶推进、航空航天、可再生能源和配电等关键能源转换应用领域必不可少的子系统。它们是由电气、机械和热管理子系统和组件组成的复杂组件。优化系统设计需要了解组件之间的电、电磁和热相互作用——在系统制造过程中，组件集成的方式对系统性能和使用寿命有重大影响。可以从组件传递到系统制造商的计算机模型需要允许有效的数字系统设计优化。电力电子元件制造商提供的现有模型仅限于电路模型，无法解释3D系统几何形状，组件放置或使用的制造工艺。可以提供三维CAD组件模型，但要发挥作用，需要详细和高分辨率的模型，这将暴露IP。然后需要复杂的有限元模拟来评估这些模型，这些模拟在计算上非常昂贵-可能需要几天才能完成。从历史上看，模型一直用于评估给定预期操作条件下最坏情况下的电气和热性能，但寿命和可靠性越来越受到关注。预测最坏情况下的电气和热性能很简单，因为最大功率和环境温度工作点可以很容易地定义和模拟。由于部件磨损是由正常工作条件下的累积应力和持续损坏决定的，因此预测部件的使用寿命和使用间隔更加困难——在可接受的性能范围内的不同条件可能会产生截然不同的使用寿命。磨损也会在很长一段时间内发生，这就需要长时间的模拟，如果使用的模型不是非常有效，那么这将进一步增加运行模拟所需的时间。所进行的研究将提出一种新的电力电子元件实时虚拟样机（RTVP）模型体系结构。RTVP模型利用降阶建模算法，使模型的模拟速度比传统有限元模型快1000倍以上。然后可以将这些模型耦合在一起并非常快速地进行模拟（在某些情况下比实时更快），以允许系统制造商在较长时间内评估系统性能，包括磨损和可靠性。此外，这些模型可以配置为隐藏敏感的设计和性能数据，这将使组件制造商能够发布准确的3D模型仿真模型，同时保护敏感的IP。这些模型可以结合起来产生系统设计的全数字“虚拟原型”，消除了对物理原型的构建和测试的需要，从而降低了设计成本并提高了系统性能。

项目成果

Datasheet Based SiC MOSFET Models for Accurate Switching Waveform Prediction in Virtual Prototyping Applications

基于数据表的 SiC MOSFET 模型，可在虚拟原型应用中准确预测开关波形

• DOI: 10.1109/dmc58182.2023.10412452
• 发表时间: 2023
• 作者: Zhang Z