Abstract: With the construction and development of new power systems, the load connotation of low-voltage distribution networks is becoming increasingly extensive, and the load fluctuations are more irregular. It is essential to fully mine the hidden value of the massive measurement data on the demand side. To quickly and accurately identify the load evolution situation, a novel idea for a data-driven, simple and efficient analysis of the load fluctuation characteristics of the low-voltage distribution network is proposed based on the concept of the motif. For the individual loads, the relative relation of load power in a continuous time period is used as the criterion for the individual motif construction, and the detailed characteristics of load variations are extracted to identify the attribute of load; for the load clusters, the relative distance of overall load power fluctuations in a continuous-time period is adopted to construct clustered motifs, which can describe the aggregated fluctuation characteristics of the dispersed load, and infer the fluctuation rules of load clusters under the influence of external factors. Numerical experiments based on the multiple actual datasets show that the proposed method can achieve the accurate cognition of load fluctuation characteristics of low-voltage distribution networks at the individual and cluster levels, respectively.