Visible light positioning method based on improved extreme learning machine

The invention discloses a visible light positioning method based on an improved extreme learning machine. The problem that an existing visible light positioning method cannot meet the high-precision real-time positioning requirement is mainly solved. According to the implementation scheme, the method comprises the following steps: acquiring RSS data from a receiving end PD and preprocessing the RSS data; constructing an extreme learning machine network ELM; updating the input layer weight and the hidden layer deviation of the ELM network; the updated parameters are substituted into an ELM network for training, and a hidden layer output matrix H is obtained; calculating the output weight of the extreme learning machine network according to the H to obtain a complete ELM network; the unknown input vector corresponding to the output vector is input into the complete ELM network after dimensionality reduction, and the corresponding output vector is the estimated coordinate of the receiving end PD. According to th