The monitoring of hole edge cracks in metal structures with holes is of great significance for ensuring flight safety and enhancing aircraft reliability. In order to monitor the crack propagation length at the hole edge, the fatigue loading test was carried out on the aluminum alloy plate with hole edge pre-cracks, and the relationship curve between the crack length at the hole edge and fatigue loading times (a-N curve) and the data of the center wavelength of the fiber grating strain sensor changing with the crack propagation at the hole edge were obtained. The method of judging the crack propagation and penetration at the preformed corner by using the center wavelength data was put forward. Then, the envelope of the collected central wavelength data is extracted by extremum method, cubic spline interpolation method and wavelet analysis, and the data envelope is obtained. Finally, the relationship between the data envelope and the a-N curve is analyzed by BP neural network, and a monitoring model that can calculate the crack propagation length at the hole edge through the data collected by the fiber grating strain sensor is obtained, which lays a foundation for the health monitoring of the aircraft structure in the future.