To solve the problem of accurate parameter identification of wind farms, an approach for identifying multi-machine equivalent parameters of wind farms based on an improved chaotic cuckoo search algorithm (CCSA) is proposed. First, the system structure of wind farms and its multi-machine equivalent modeling are described, and an improved scheme of the CCSA is designed. Then, by combining the analytical and identification methods, a sequential identification process of multi-machine equivalent parameters of wind farms based on the improved CCSA is established. Finally, a simulation model of the wind farms with multiple doubly-fed induction generators (DFIGs) is built using the Matlab platform, and the effects of the proposed sequential and simultaneous identification methods in identifying the wind farm parameters are compared. The performance differences among the improved CCSA, traditional particle swarm optimization (PSO), and cuckoo search algorithms (CSA) are analyzed. The simulation results show that the proposed identification method can reduce the DFIG’s parameter identification average error from 11.07% to 2.41%, and more precisely imitate the dynamic characteristics of the wind farms. Therefore, the effectiveness of the proposed approach is well validated.