Abstract: The power load of the smart grid fluctuates increasingly high and rapidly because of the introduction of intermittent energy, such as wind energy and solar energy. As a result, the increase of the randomly fluctuating load in the smart grid brings new challenges to the active power measurement of smart electricity meters. However, the installed power meter and the standard electricity meter are designed for a steady input signal. The traditional MA (moving average) and ⅡR (infinite impulse response) measurement algorithms are proposed for the steady situation and are thus not suitable to address the dynamic error testing and metering problems. Moreover, although some harmonic experiments have been performed that provide an overview of domestic and internal standards of electricity energy meter, there is a lack dynamic characteristics in them. Thus, it is of great theoretical significance and application value to study the dynamic measurement characteristics of the existing smart electricity meters and propose an effective dynamic measurement to improve the metering dynamic performance. To reduce the measurement error of a smart electricity meter under dynamic load power conditions, a SDPA algorithm for dynamic active energy measurement was proposed in this work. First, the dynamic response speed and dynamic error characteristics of active power of traditional MA and ⅡR low pass filter algorithm were deduced. Next, the limitations of the two algorithms for dynamic input signal were highlighted, and the influence factors were determined by theoretical analysis. Based on these results, a SDPA algorithm for dynamic measurement of smart electricity meter was proposed. The new algorithm was implemented by truncating periodically, executing piecewise point product operation and summing up the active power. In addition, the implementation method by decimation can save storage space and improve the operation speed. The theoretical and simulation results show that the SDPA algorithm can reach a lower error level in one period of response time.