The invention provides a building thermal load hybrid prediction method and system based on periodic data driving, and the method comprises the steps: carrying out recognition of periodic characteristics of collected building thermal load data through a Fourier decomposition method, and obtaining a building thermal load data residual error after the periodic characteristics are removed; training and optimizing the key parameters of the building thermal load hybrid prediction model by using the data residual error and adopting a particle swarm evolution algorithm; utilizing the trained and optimized building thermal load hybrid prediction model to respectively predict the thermal load disturbance quantity of the personnel thermal behavior in the building, the thermal load loss quantity caused by indoor/outdoor temperature and humidity changes and the building thermal load prediction value; fusing the predicted thermal load disturbance quantity, thermal load loss quantity and building thermal load prediction value to obtain a building thermal load comprehensive prediction value; according to the invention, comprehensive and accurate prediction of the building thermal load is realizedby using the building thermal load data with multiple complex characteristics, and the comprehensive management and control capability of the building thermal load is improved.