Improved RBF neural network indoor visible light positioning method and system

Abstract

The invention discloses an improved RBF neural network indoor visible light positioning method and system. An LED in an indoor space emits a light signal, the light signal passes through an indoor space channel, and a receiving end converts the obtained light signal into an electric signal for positioning through a PD detector; and the method comprises the steps of calculating an illumination intensity value with minimum optical power fluctuation, calculating positioning, and optimizing an RBF neural network; the optimization of the RBF neural network comprises the following steps: improving the performance of the RBF neural network by using a KPCA-K-means++ model, namely clustering RSSI data collected by a receiving end by using the KPCA-K-means++ model to obtain an optimal clustering center and an optimal clustering number as the number of neurons and a neuron center of a hidden layer; and establishing a GA-LMS model to optimize RBF neural network parameters, that is, obtaining accurate width and connection weight by utilizing GA-LMS, and finally obtaining accurate positioning coordinates , so that the method is higher in generalization.

Description

technical field [0001] The invention relates to an improved RBF neural network indoor visible light positioning method and system based on a KPCA-K-means++ model and a GA-LMS model. Background technique [0002] With the development of positioning technology becoming more and more mature, people are no longer satisfied with positioning in an outdoor environment, and indoor positioning technology has gradually attracted the attention of experts. If we can make full use of indoor positioning technology, then this technology will bring more convenience to our life and can solve the location problems in various indoor places. The Global Positioning System (GPS) is widely used in positioning technology, but due to the attenuation of the satellite signal reaching the ground, it is difficult to penetrate a complex building, which will cause a large positioning error [1] . Existing indoor positioning technologies include infrared indoor positioning technology, Bluetooth indoor pos...

AI Technical Summary

Problems solved by technology

However, there are many problems in RSSI positioning that will cause positioning errors: First, due to indoor light shading, people walking and other problems, the received optical power of each corner of the indoor space is not the same, and the positioning accuracy of each corner of the indoor space is different.

Second, the least squares in the RSSI positioning algorithm will produce ranging errors and other problems, which will lead to reduced positioning accuracy and inaccurate positioning

Due to the different intensity of the optical signal received by the PD and the uneven distribution of indoor light, it is impossible to accurately find the location indoors.

Images