The effectiveness of parameter identification is strongly related to coupling errors, which exhibit highly complexity characteristics. However, traditional particle swarm optimization (PSO) algorithms are difficult to efficiently utilize the coupling errors to expand their search range. This paper proposes an affine parameter identification method for modular multilevel converter static synchronous compensator (MMC-STATCOM) based on an improved PSO. First, the mathematical model of the MMC-STATCOM is established to determine the parameters to be identified. Next, an affine-based parameter identification method is proposed to transform the parameter identification problem into a joint identification problem of parameters and coupling errors. Building on this, the entropy weight method is adopted to comprehensively consider the influence of interval satisfaction and interval errors on the identification results. The improved PSO is then used to realize decoupling identification of the parameters and coupling errors for the MMC-STATCOM. Simulation results prove that the identification errors of parameters and coupling errors are within 1.06% and 2.95%, respectively.