Most identification methods can only be applied to closed loop parameter identification by specific input signals. In order to solve the, parameter estimation problem of a closed loop system with different test signals, a novel improved optimization method called the swarm cuckoo search is proposed. The swarm cuckoo search algorithm adopts a special approach of computing discovered probability, and it is different from other cuckoo search algorithms. The proposed algorithm has a strong ability to locate the global minimums with random initial values ins the search range. Additionally, simulations also indicate that the proposed algorithm can increase the accuracy of the parameters when compared with particle swarm optimization algorithm.