As a dominant form of onshore wind power supporting China’s twin goals of carbon peak and carbon neutrality goals

large-scale doubly-fed induction generator (DFIG) wind farms with high-penetration power electronic devices impose increasingly stringent requirements on electromagnetic transient simulation technologies for power systems. However

research on full-topology refined microsecond-level simulation for DFIG-based wind farms with hundreds of generation units remains relatively scarce. This paper proposes a parallel simulation methodology for DFIG wind farm clusters based on field programmable gate array (FPGA)

utilizing their microsecond-level small time-step computation capability. Firstly

a microsecond-level highly parallel discrete model is established for doubly-fed induction machines

accompanied by controlled-source modeling development for power converters. Subsequently

the nodal admittance matrix is partitioned and reduced in dimension to achieve internal network parallelism within the power generation unit

and a unit-level parallel simulation framework is simultaneously constructed at the global circuit-solving level. Finally

considering the characteristics of real time digital simulator (RTDS) and FPGA platforms

microsecond-level small-step simulation tasks are allocated for generation units while establishing a co-simulation hardware architecture through platform- optimized task distribution. Through comparative analysis between RTDS benchmark models and RTDS+FPGA co-simulation models

the accuracy of the proposed parallel simulation method is systematically validated.