Abstract: In view of the low accuracy of dynamic prediction of mining subsidence and less research on the dynamic prediction of inclined main section,in order to improve the prediction accuracy and the pertinence of dynamic prediction,and strengthen the guidance role of dynamic prediction results in the production of mining area and the land use of goaf,this paper,based on the static prediction model of probability integral,discusses the prediction method of dynamic subsidence for inclined main section by using the “optimized segmented Knothe time function”.The main contents are as follows:firstly,the division method of dynamic mining units for inclined main section is discussed.Secondly,based on the limited mining principle of probability integral method,the parameter calculation formula of each dynamic unit is derived.Thirdly,according to the superposition calculation principle,combining the time function value of each unit,a dynamic prediction model for inclined main section,which is also suitable for horizontal and gently inclined coal seams and rectangular and approximate rectangular working surfaces is established.Fourthly,according to the established model,the calculation process of dynamic prediction is discussed,the corresponding computer programming algorithm is developed,and the difficulties to be noticed are analyzed.Finally,according to the model and algorithm in this paper,the dynamic prediction program is compiled and applied to the prediction practice.Prediction practice 1 shows that no matter which direction the mining begins from,the direction of uphill or downhill the prediction,the results of dynamic subsidence and deformation of inclined main section are always consistent with the theory rules.In addition,when the given prediction time is long enough,the prediction result of dynamic subsidence is also consistent with that of static prediction.Therefore,the scientific basis of the model and algorithm is revealed theoretically,and the integration of dynamic and static prediction of mining subsidence is achieved.Prediction practice 2 shows that it can be seen from the comparative analysis of the dynamic prediction results and the measured results of the 29401 working face,the maximum mean square error of prediction is 452 mm and the minimum is 54 mm.On the whole,the relative error of prediction can be limited to about 8.5%,thus,the reliability of the model and algorithm is proved.