Deep learning-based Wi-Fi fingerprint reconstruction method and system for spatial positioning

The invention discloses a Wi-Fi fingerprint reconstruction method and system based on deep learning for spatial positioning. Comprising the steps that the mobile robot collects and uploads data; performing background processing on the data; an overall framework is formed by a progressive reconstruction network for non-uniform fingerprint map reconstruction, a position coding optimization module for repairing the network and a multi-feature dynamic optimization module for optimizing the network, an original sparse fingerprint map is reconstructed into a dense fingerprint map, and a learning model is generated through neural network training. Outputting a high-precision fingerprint map; and obtaining a dense fingerprint map for subsequent space positioning. The fingerprint map reconstruction network overcomes the problems that in the prior art, when signals are not uniformly distributed, the fingerprint map reconstruction effect is not ideal, part of position information is lost in the reconstruction process, t