The load model has an important influence on the simulation analysis results of the power system, but the current load modeling methods are difficult to adapt to the diversity and time-varying characteristics of the new power system load in terms of model accuracy and computational efficiency. To address this issue, an online modeling method of the comprehensive load model for the new power system based on measurement information is proposed. Firstly, the dominant factors affecting the comprehensive load model are analyzed based on the activation subspace of parameter space. Then, according to the characteristics of the daily output curve of the load, a large number of underlying loads are clustered by using a clustering algorithm. Finally, the underlying loads are equivalently aggregated from the low voltage level to the high voltage level, and the equivalent aggregation model of the comprehensive load is constructed. The analysis results of the example show that the load model established by the proposed method can better track and record the real-time changes of the system load and more accurately reflect the actual load characteristics.