Physical-data fusion power vacancy prediction method and system considering load inertia

The invention provides a physical-data fusion power vacancy prediction method and system considering load inertia, and relates to the technical field of power vacancy prediction. Comprising the following steps: establishing an inertia center frequency model, and updating an inertia center frequency curve and the overall inertia of each generator on a power generation side; adopting a sliding time window to select a plurality of data points on the updated inertia center frequency curve, fitting the data points to obtain a local curve of the corresponding time window, and taking a tangent slope of a midpoint of the local curve of the time window as a frequency change rate of the corresponding time window; a time window corresponding to the optimal frequency change rate is selected, and unbalanced power borne by the power generation side generator set is predicted; predicting unbalanced power borne by the load side based on a pre-trained deep learning network model; and adding the unbalanced power borne by the p