Research On Visible Light Indoor Positioning System Based On Neural Network

Visible light indoor positioning technology is a new type of indoor positioning technology,which organically combines lighting and wireless positioning.Compared with traditional indoor wireless positioning technology,it has huge development prospects.However,the existing visible light indoor positioning technology is difficult to overcome the interference of background noise and indoor reflected noise,resulting in a decrease in system positioning accuracy.Aiming at the problem that the visible light indoor positioning system based on RSS is susceptible to background noise and indoor reflection noise,this thesis introduces the BP neural network into the RSS positioning algorithm,and proposes a visible light wireless indoor positioning model that combines the neural network and RSS.Firstly,a visible indoor stereo positioning system model is built.To further reduce the impact of diffuse reflection of optical channel on positioning accuracy,a channel mathematical model including direct line-of-sight link and first-order reflection link is established.Secondly,after determining the LED light source layout model,combined with the LED channel diffuse reflection model,grid calibration is performed on the receiving planes of different heights.At different positions on different planes,the light intensity of different LEDs at the center points of each grid is collected,and the collected data is classified and processed,and then a visible light signal training data set and a test data set are created respectively.Finally,the indoor positioning error model algorithm based on the fusion of BP network and RSS is designed,and the data is trained and predicted.The space vectors of the distance between the position coordinates of the predicted receiver and the position coordinates of different LED light sources are obtained,which can be used for rough positioning of the target location to be measured.The location error constraint model based on the combination of positioning variance and Euclidean distance is used to determine the specific location of the target to be measured step by step.In this thesis,the theoretical validity of the algorithm and the reliability of positioning are carried out respectively in simulated positioning experiments and actual positioning experiments.In the process of simulation positioning test,this thesis uses a three-dimensional space with a length of 4 meters,a width of 4 meters and a height of 3 meters as a model,and the simulation experiments are carried out on three height planes.From the positioning experiment data,the average positioning error is 1.67 cm.In the actual positioning stage,an experimental platform was built in a three-dimensional space with a length of 0.8 meters each,and multiple positioning test experiments were performed,and the average positioning error was 3.65 cm.