In view of the common noise interference in the signal acquisition of power system, an adaptive morphological filter using composite structure elements is proposed. The adaptive filter combines two or more structure elements as the composite structure element. The filtering error between the input and output signals is analyzed to determine the integral size of the composite structure element that can make the filtering effect reach the optimum. Then the parameters of the two elements of composite structure elements are adjusted according to the filtering error maximum principle for further optimization. Finally the optimal composite structure element is obtained. Simulation experiment is carried out on the adaptive filtering of the power signal containing random white noise. The results show that the proposed adaptive filter can accurately find the optimal structure element in the face of random noise. Compared with the traditional morphological filter using a single structure element, it has better filtering performance and also has a good application prospects. This work is supported by National Natural Science Foundation of China (No. 51277066).