In agricultural production, it is challenging to detect the flow rate of fertilization due to the large amount and the mutual shielding of fertilizer particles. A granular fertilizer flow detection method based on the principle of luminous flux blocking was proposed to solve this problem. The feasibility of the detection method was preliminarily determined by theoretical analysis; discrete element simulation was used to analyze the distribution of granular fertilizer discharge in the conveying tube, which provided a basis for the structural design and installation position of the particle flow detection sensor (PFDS). Based on the above analysis, the PFDS and testbench were designed. Experiments were carried out by selecting urea and compound fertilizer as materials and the fertilizer discharging wheel (FDW) speed as factors to verify the method. The results showed a strong linear correlation between fertilizer flow and the cumulative response voltage of PFDS. Under different fertilization frequencies, the correlation determination coefficient (R2) of both was higher than 0.992. The detection models of each fertilizer frequency were established. Absolute percentage error (AE) and mean absolute percentage error (MAPE) were selected as the index. The MAPE of the detection model based on the “accelerate” for urea and compound fertilizer were 5.18% and 4.07%, respectively. And the detection error was lower than that of others, so the establishment method of the optimal detection model was determined. To solve the matching problem between the PFDS and fertilizer tube with different diameters, the experiments were carried out by taking the number of sensitive elements and the inner diameter of the PFDS as factors. The results showed that when the density of detection elements was 0.075~0.75, the MAPE range for urea and compound fertilizer were 4.75%~9.33% and 4.07%~9.11%, respectively, and the MAPE was decreased with the increase of detection element density. The detection method would provide technical support and solutions for granular fertilizer flow detection.