End-to-end Design Automation for Reliable Power System Components

Surrogate modeling is an integral part of power systems operations, planning, diagnostics and design. Satisfactory training of surrogate models relies on sufficient amount of high-quality representative data. In surrogate modeling that requires complex engineering design optimization across multiple domains, a critical challenge is to scale the designs to a wide range of parameters. Such a scaling needs to satisfy the performance requirements and ensure robustness against manufacturing variations, environmental stresses, and aging effects. The current approach in many engineering fields involves iterative, time-consuming processes that rely heavily on domain experts conducting repetitive simulations with minor parameter adjustments. Achieving reliable operation under extreme conditions,…