With the accelerating urbanization, the load demand for urban power networks is swiftly increasing. Accurate prediction of changes in the power grid is extremely important to ensure stable operation, planning, and power system construction. First, this paper analyzes the load characteristics of urban power grids, including the effects of seasonal changes, daily changes, business days, and non-working days. Second, a load forecasting model is built using time series analysis, regression analysis, and machine learning. The back-test verification of the urban power grid data demonstrates that the model has achieved high prediction accuracy. Finally, this paper predicts the future development trend of power grid load, which provides a reference for the optimal operation and investment decision-making of the power grid. Various data analyses and models are adopted to analyze the complexity of urban power grid load characteristics. Specifically, time series analysis is a traditional method to deal with such problems. It can identify the periodicity and trend of data and predict accordingly. In addition, regression analysis can reveal the correlation between load and diversified factors. With the development of artificial intelligence technologies, machine learning demonstrates the great potential of load prediction, particularly deep learning techniques, which is good for dealing with big data and nonlinear issues.
Yi Zhang (2025) studied this question.