In the real practice applications, classical composite load model (CLM) is widely used and developed in the power grid. In this paper, CLM considering frequency characteristics was interested in. With instantaneous sampling data of alternating current voltage and current, the CLM's parameters were identified by load parameter identification strategy based on measurement-based method ant colony algorithm with stronger capability of global optimization. Through simulation and calculation, the impact of sampling data with different length of time on parameter identification results of load model was comparatively analyzed. The impact of sampling data with the same time length on the curve fitting results of load model was also analyzed.