Abstract: With the acceleration of the construction of new power systems, the large-scale application of renewable energy and nonlinear dynamic loads has caused serious deviations in electricity metering. To study the important features that cause measurement errors in complex power dynamic load signals, this paper proposes a typical feature extraction and feature modeling method based on waveform domain and run length domain. Firstly, on-site collection of large-scale complex power dynamic load signals from electrified railways and electric arc furnaces is carried out, and a discrete mathematical model is constructed to analyze their typical features in the waveform domain and extract feature parameters; Secondly, the mapping method of signals from waveform domain to run domain is studied, and run domain feature parameters are constructed to characterize the global characteristics of rapid and random dynamic fluctuations in load current; Finally, based on the typical characteristics of the run domain and waveform domain of complex power dynamic load signals, constraint conditions were constructed. Using feature modeling methods, a binary dynamic current test signal model with specific parameters was constructed, and the experimental analysis showed that the test signal can reflect the impact of typical dynamic load characteristics on energy metering errors, which is effective.