WiFi signals are unstable in complex indoor environments and the distortion effects of buildings on the geomagnetic field results in the low accuracy of single-source positioning. Considering this, this study adopts multi-source information fusion positioning technology that can effectively use WiFi and fingerprint data of the geomagnetic field for positioning and proposes an improved adaptive differential evolution algorithm to optimize the BP neural network (IDEBP). This method optimizes the weights and deviations of the BP neural network by improving the mutation, crossover, and selection operation of the differential evolution algorithm, which helps the BP model to better learn the characteristics of WiFi and fingerprint data of the geomagnetic field. The simulation shows that the proposed algorithm greatly improves the accuracy of indoor fingerprint positioning.