As location-based service technology continues to develop, people's demand for positioning technology is increasing. Satellite positioning technology can be widely used for positioning in outdoor environments. However, due to the weak penetration ability of satellite signals into buildings, the application of this technology in indoor environments faces many limitations. Therefore, there is an urgent need to conduct in-depth research on reliable indoor positioning technology. Among commonly used indoor positioning technologies, indoor positioning technology based on WLAN signals has the advantages of wide coverage, low deployment cost, easy installation, and non-line-of-sight propagation, so it shows excellent universality. This article proposes a Wi-Fi Positioning Method Based on convolutional autoencoder and device heterogeneity compensation. Compared with traditional Wi-Fi positioning methods such as SVM, the positioning accuracy has been greatly improved and can meet daily low-cost, high-precision indoor positioning needs.