Indoor positioning technology as the urgent demand of various industries of science and technology services, there is no recognized perfect solution. Because every positioning technology can not eliminate its inherent shortcomings, the integration of multiple positioning technologies is an important research direction to achieve high precision indoor positioning. Facing increasingly complex indoor environment, a multi-source indoor positioning method is proposed in the paper, which combines deep confidence network and RSSI fingerprint positioning method to achieve rough positioning. At the same time, pedestrian position measurement technology is used to complete pedestrian track prediction. Then, the particle filter is used to fuse the rough positioning results with the predicted pedestrian track information, which improves the accuracy and real-time performance of the traditional RSSI indoor fingerprint positioning technology.