Analysis and Forecasting for Power Load of Office Buildings Taking Crowd Behavior into Account

With the advancement of big data technology and sensor technology, big data of power systems is gradually being established. Specialized power load analysis and prediction for different types of users has become an emerging hot spot. However, there is still little research on the power load of office buildings and crowd behavior has not been taken into account in the existing power load analysis. What’s more, existing correlation analysis methods are not well adapted to the multi-source heterogeneity of big data. Aiming at the above problems, a comprehensive association analysis method for multi-source heterogeneous data is proposed. Taking the mobile base station data as the characteristic data of the crowd behavior, the relationship between the office building power load and its influencing factors is tapped by the above method. Finally, taking the measured data of two office buildings in Shanghai as an example, mathematical analysis for office building load through statistical and clustering methods and short-term load forecasting is carried out by means of power load correlation. The results show that the analysis method has strong robustness, and considering the crowd behavior can also improve the accuracy of short-term forecasting for power load of office buildings.