<p>In order to enhance the warning efficiency of the overload operation status of the main transformer and improve the real-time response capability of the warning, this paper proposes a real-time warning algorithm for the overload operation status of the main transformer based on multi-source data-driven approach. Focusing on the overload operation scenario of the main transformer, the data collection items are accurately determined, and data information covering multiple sources are widely collected. Construct a dataset of the overload operation status of the main transformer using the collected data, deeply analyze and determine the key features that can reflect the overload operation status, and reasonably complete the selection of labels. The Yang Hui triangulation method is used to smooth the data, combined with data cleaning techniques, to accurately fill in the missing features in the state data and effectively eliminate redundant data, thus constructing a high-quality multi-source dataset of the overload operation status of the main transformer. Based on linear regression algorithm, establish a model of the overload operation state of the main transformer, determine the weight of the overload operation state through state combination weighting method, and then calculate the overall deterioration degree of the overload operation state of the main transformer, and achieve real-time warning based on the deterioration degree. The experimental results show that when the sample size reaches 5&#xa0;k, the method proposed in this paper performs excellently in the warning of overload operation status of the main transformer, with a warning accuracy rate of 97.59%, a warning accuracy rate of 96.3%, and a warning recall rate of 89.6%. This series of data fully demonstrates that the method proposed in this paper can effectively improve the warning effect of the overload operation status of the main transformer.</p>