A grid division-based detection and anti-lost method for the abnormal trajectory of the elderly. Firstly, the map is divided into a grid matrix, that is, a parameter D is set to divide the map into square grids. Then, for the trajectory data uploaded by the elderly fall alarm, after denoising processing, the trajectory data is mapped to the corresponding grid, and it is judged whether the elderly appear in the historical activity grid within T time. If so, refresh T and continue processing the trajectory stream. If it does not appear, an alarm message will be pushed to the family through the APP, and the family will determine whether it is indeed lost. If it is indeed lost, it will navigate to the location of the elderly. If not, refresh T and add the grid that the elderly passed through at this stage to the history. In the active grid, the next time it appears in this grid, the alarm will not be triggered, and then continue to process the estimated data flow information. This method has three main advantages: first, it can use historical trajectories to automatically delineate the range of safe activities; second, it can dynamically adjust the range of safe activities according to system operation; Abnormal trajectory to the police to the family.