Recent studies have shown that the power and current of high-power dynamic loads exhibit dynamic behaviors that cause large errors in electrical energy meters. Aimed at fast variations and large fluctuations in dynamic electric signals, this study examines a characteristic extraction method for small-time granularity and the impact of these characteristics on electrical energy meters. First, we constructed quasi-steady and dynamic model parameters for the dual-modal modulation models of current and power signals. Second, based on the two sets of signals sampled from the electrified railway traction substation and steelmaking arc furnace substation, we extracted quasi-steady and dynamic signal model parameters from fundamental active power signals. Third, we proposed three novel characteristic parameters by mapping the model parameter method, and extracted the four important characteristics. Finally, we designed a set of on-off-keying test signal experiments to determine the impact of the proposed characteristic parameters on the electric energy meter. The experimental results show that the electrical energy meter errors are sensitive to the proposed characteristic parameters, wherein the maximum impulse coefficient is the most sensitive characteristic of the EEM, which causes an error as large as −81.28%. The proposed characteristic parameters can be used as the input characteristic information of dynamic test signals for type testing and updating the test standards for electrical energy meters.