Two-Level Surrogate-Assisted Transient Parameters Design Optimization of a Wound-Field Synchronous Machine

Abstract

The rapid development of ultra-high voltage direct current transmission systems has placed greater demands on the design optimization of transient parameters for large wound-field synchronous machines (WFSMs) in power systems, such as large synchronous condenser for dynamic reactive power compensation. This paper proposes a novel two-level surrogate-assisted multi-objective optimization method for WFSMs that can optimize transient parameters including per-unit reactance and time constants rapidly and accurately. To facilitate the manufacture of a prototype for verification, the proposed method is applied to the optimization of a WFSM for scale-down generator application. This method is employed to deal with the excessive number of design parameters, issues of multi-objective evaluation and strong coupling between objectives that either result in inaccuracy or become time-consuming by using conventional methods. To generate the training set for the surrogate model efficiently by finite element analysis, an improved arbitrary rotor position standstill time response (SSTR) method is employed to identify all transient parameters simultaneously in a single test, while Latin hypercube design is used to capture data features. The optimal WFSM designed by using the proposed method is prototyped and measured. It confirms that the proposed method can achieve high validity and high optimization efficiency simultaneously.