Multiscale Analysis of Regional Power Load and Its Associations with Meteorological and Socioeconomic Factors: Evidence from Shandong and Inner Mongolia

Regional power load is an important basis for renewable energy accommodation, power grid capacity planning, and demand-side resource management. Its variation is jointly associated with meteorological conditions and industrial and consumption-related conditions, which may operate at different temporal scales. To address this issue, this study takes Shandong and Inner Mongolia as study areas and develops an hourly–monthly–annual multiscale analytical framework. Variational mode decomposition is first used to identify low-frequency trends, intraday cycles, intra-annual fluctuations, and high-frequency disturbances in load series. Pearson correlation and mutual information are then applied to characterize linear associations and more general statistical dependencies between load modes and meteorological, industrial, and consumption factors. Annual maximum load, annual electricity consumption, meteorological variables, and key socioeconomic indicators are further examined to describe the long-term regional backgrounds of load variation. The results show that (1) both regions exhibit evident multiscale load structures, and external-factor associations are concentrated in different load modes; (2) separate within-region hourly analyses show that meteorological associations in Shandong are mainly concentrated in the 24 h daily-cycle mode, particularly for radiation- and sunshine-related variables. In Inner Mongolia, meteorological associations are mainly observed in the low-frequency trend and daily-cycle modes; (3) at the monthly scale, meteorological associations in both regions are mainly distributed in the annual and intra-annual components, and many corresponding Pearson associations were retained by the unadjusted block-bootstrap analysis, whereas the socioeconomic associations retained by the same analysis were relatively sparse and localized; and (4) at the annual scale, meteorological and socioeconomic trajectories are presented only as descriptive regional context because only seven observations are available for each region. These findings indicate that decomposition can help identify the temporal components in which overall load–factor associations are concentrated and provide a descriptive basis for subsequent component-specific variable screening and regional load analysis.