The invention relates to a dynamic model based method for realizing secondary compensating control when a numerical control system passes through a quadrant. The method comprises the following steps of enabling the numerical control system to seek for a compensation starting point, a compensation terminal point and a compensation maximum point of a position single-axis error in a position passing through the quadrant; structuring a compensation sequence, establishing a first dynamic compensation model between the compensation starting point and the compensation maximum point, and establishing a second dynamic compensation model between the compensation maximum point and the compensation terminal point; calculating an error between the first dynamic compensation model and an actual position error, as well as an error between the second dynamic compensation model and the actual position error, so as to obtain area distribution enabling the first dynamic compensation model and the second dynamic compensation model to be most attached to the actual position error, and according to the region, performing actual pulse compensation which can be performed repeatedly. Through the adoption of the compensating control method adopting the structure disclosed by the invention, under the self-adapting condition, the secondary compensation can be performed on the model, so that the whole reduction of the position error in the position passing through the quadrant is completed, the best compensation attaching can be realized, the accuracy of the numerical control system is improved, and the operation is convenient.