Multi-carrier dam damage identification method and device based on weight self-adaption

The invention discloses a multi-carrier dam damage identification method and device based on weight self-adaption, and the method comprises the steps: carrying out the recognition and detection of the damage of a dam caused by the natural disasters, such as an earthquake, through the aggregation of a plurality of unmanned carriers through a self-adaption weight calculation and model layering mixing technology; an adaptive damage identification model (namely, a personalized model) is provided for each unmanned vehicle, and the damage identification accuracy of the unmanned vehicle is ensured at the same time. The self-adaptive weight calculation is used for updating a layered weight value through a chain derivation rule according to model parameters on each carrier, so as to obtain an optimal mixed weight of each carrier model; hierarchical mixing is carried out on the intermediate versions (namely intermediate models) of the local models and the personalized models on the carriers by using the hierarchical mi