A Data-Driven Method for Extracting Temporal Characteristics of Load in Typical Industrial and Commercial Users

With the continuous expansion of the scale of industrial and commercial loads, the load timing after its connection to the power grid presents the characteristics of complexity and diversity. Traditional load analysis methods are difficult to integrate multi-source data and adapt to dynamic scenarios, and it is difficult to meet the requirements of power grid stability for load analysis. In this context, this paper proposes a data-driven method to extract typical industrial and commercial users' load time series features. Firstly, a multi-level load feature database is constructed through artificial prior information and machine learning algorithm. Then, the abnormal data is cleaned and corrected, and the accurate data classification is realized by improving the support vector machine (ISVM) model. Finally, 90 days of historical load data of typical equipment were selected for simulation. The results show that the average classification accuracy of the proposed method is 89.52%, which can effectively identify the user load regulation potential and provide theoretical support for demand response and energy efficiency optimization.