A Typical Day-iTransformer Based Method for Power Load Breakpoint Fitting

Amid the continuous advancement of China's electricity market reform, accurate user - side power load aggregation monitoring is of great significance. This paper focuses on the breakpoint issue in load aggregation curves. By leveraging the Typical Day - iTransformer algorithm, a feature set is constructed, data features are mined, and breakpoint fitting is effectively achieved. Comparative analysis reveals that this algorithm outperforms other methods, ensuring data integrity and offering strong support for subsequent power - related applications. Through the verification of single - point and multi - point missing scenarios, it is found that the Typical Day - iTransformer algorithm exhibits significantly better fitting results than the Newton interpolation algorithm when there are missing values in the load curve, providing a new solution for optimizing curve numerical fitting.