Convolutional Neural Network based UWB/BLE/BDS Fusion Positioning System

Aiming at the problem that it is difficult to realize the seamless positioning in the complex indoor and outdoor positioning environment, a fusion positioning algorithm based on convolutional neural network is proposed. Using its strong ability of image processing, the target positioning is obtained by transforming the positioning data fusion processing into solving the image classification, and the algorithm is applied to the UWB/BLE/BDS seamless positioning system. In the transition region, the positioning points obtained from a single subsystem are filtered first to form a coordinate map, which is input into the convolution neural network with supervised learning for training, subsequently, the test data is processed and input into the trained network to obtain the fusion weights based on convolution neural network in the actual measurement stage, computing out the positioning coordinates finally. The experimental results show that the average positioning accuracy of the system in the transition region processed by the weighted fusion algorithm based on convolution neural network is improved by 42.9% than that of the average weighted fusion algorithm.