Propagation of uncertainties across electromagnetic models

在电机的生产过程中，由于生产过程的不同，所使用的材料及其性能会发生偏差。这些影响的考虑通常是通过最坏情况估计和蒙特卡罗模拟来执行的。为了提高估算的精确度和减少计算量，本文提出的项目目标是“电磁模型中不确定因素的传播”精确化项目的结果传递。从电机的一个数值模型的有限元分析开始，设计一个元模型。该元模型将用于研究材料性能偏差对机器特性的影响。将执行故障模式和影响分析(FMEA)以估计偏差程度。在已知这些偏差的情况下，机械传递函数和声学特性将被转换为时域机器模型。随后，输入参数的敏感度分析允许设计元模型。元模型将计算时间开销较高的时间域模型转换为代价较低的多项式表面积响应。根据元模型的结果，将得出优化机械设计的建议。这包括关于恒定边界条件的优化、稳健设计、增加可靠性的建议以及关于公差分配的建议。项目合作伙伴Wilo SE以其在开发小批量和大批量电机方面的经验为该项目提供支持。该项目中要调查的电机的设计和开发也将是Wilo SE贡献的一部分。合作伙伴提供的机器将通过测量进行分析，以调查其特性。优化机器设计的建议将参考那些参考机器设计。

During the production of electrical machines, the used materials and their properties deviate as a result of the production process. The consideration of these effects is usually performed by worst-case estimates and Monte-Carlo simulations. In order to increase the accuracy and reduce the computational effort of said estimates, the hereby proposed project targets at the transfer of results of the preciding project "Propagation of uncertainties across electromagnetic models". Starting with a finite element analysis of a numerical model of the electrical machine, a meta-model will be designed. This meta-model will be used for the investigation of the influence of material property deviations on the characteristics of the machine. A Failure Mode and Effects Analysis (FMEA) will be performed to estimate the degree of the deviations. With the knowledge of these deviations, the mechanical transfer function and the acoustical characteristics will be transformed into a time domain machine model. Subsequently, a sensitivity analysis of the input parameters allows the design of the meta-model. The meta-model transforms the time domain model with high costs with respect to computation time into a polynomial surface area response with lower costs. Based on the results of the meta-model, suggestions for an optimized machine design will be derived. This includes suggestions regarding the optimization for constant boundary conditions, for robust design, for the increase of reliability as well as suggestions for tolerance allocation. The project partner WILO SE supports the project with their experience in the development of electrical machines in small and big quantities. The design and development of the electrical machines to be investigated in this project will also be part of WILO SE contribution. The machines provided by the partner will be analyzed by measurements in order to investigate their characteristics. The suggestions for optimized machine designs will refer to those reference machine designs.