This paper focuses on the windmill starting performance of intercooler turbofan engines and develops a simulation model based on component characteristics. The results show that the engine’s windmill performance is significantly influenced by the flight Mach number and altitude. As Mach number increases, the windmill speed and resistance rise, while the bypass ratio decreases. With an increase in altitude, both the windmill speed and resistance decrease, but the bypass ratio remains nearly constant. During the windmill starting process, the matching state of various components is primarily affected by the flight Mach number. The starting process under different altitudes follows nearly identical working lines, though the start-up endpoint varies. To address the issue of poor convergence in the windmill state model, this paper proposes a surrogate model based on Gaussian Process Regression (GPR), which, by integrating expert knowledge, effectively improves computational efficiency and accuracy. The average error is less than 0.5%, and the maximum error is less than 4%. The research findings provide theoretical support for the design and optimization of intercooler turbofan engines and offer a new approach to solving the low-speed convergence issue in windmill starting performance studies.