Application of Laser Ranging and Grey Neural Network (GNN) for the Structure Optimization of Radio Frequency Identification (RFID) Tags

In the dynamic detection of radio frequency identification (RFID) system, the structure distribution of the tags has an important influence on the reading distance of RFID tags. In order to analyse the influence of tag structure distribution on tag reading distance, this paper proposes a method for optimal distribution of the three-dimensional (3-D) coordinate structure of the tags based on laser ranging and Grey neural network (GNN). Firstly, a dynamic performance test system based on laser ranging is adopted. Then, by using the dynamic performance test system, the reading distance of different tag 3-D structure distribution is obtained. Finally, in view of the non-linear relationship between the 3-D structure distribution of the tags and the corresponding reading distance, the GNN model is used to model this non-linear relationship. The established model is used to predict the reading distance of the tags' 3-D structure distribution. Experimental results show that compared with genetic algorithm-back propagation (GA-BP) neural network, particle swarm optimization-back propagation (PSO-BP) neural network and support vector machine (SVM), the GNN method proposed in this paper can more accurately predict the reading distances of tags' 3-D structures. The prediction error values of root mean square error (RMSE), mean absolute percentage error (MAPE) and mean absolute error (MAE) are 0.0469, 0.0258, and 0.0402, respectively. The method proposed in this paper can optimize the 3-D structure distribution of RFID tags.