Research,Design And Implementation Of An Intelligent Supermarket Management System Based On RFID And Machine Learning

With the rapid development of e-commerce,traditional supermarket ushered in a huge challenge,operating costs much higher than the shop,so seeking a low-cost mode of operation is necessary for the operators of traditional supermarket.Combining the advantages of electronic tags of non-contact automatic identification and machine learning can learn the characteristics of data changes,the presented intelligent supermarket management system will bring a huge competitive advantage to the traditional supermarket.This paper discusses the significance of the research on intelligent supermarkets and the application of RFID technology to intelligent supermarkets at home and abroad,analyzing the existing shortcomings and feasible solutions,introducing the commonly used localization algorithm and putting forward the shortcomings of these algorithms.We present two kinds of indoor positioning algorithms based on machine learning and developed the intelligent supermarket management system.The main innovations and research contents of this paper are as follows:1.In the traditional machine learning field,XGBoost algorithm with its excellent performance to make it very popular.In this paper,we present a new algorithm for indoor positioning based on XGBoost.The algorithm predicts the coordinate position of the tags by learning the relationship from the RSSI value vector of the reference position and the coordinates of the reference tags.Simulation experiments verified that the accuracy of the algorithm is 57.11%higher than that of the traditional LANDMARC algorithm,and the accuracy of the algorithm is also better than that of the LANDMARC algorithm under different reference labels.2.In the previous research,some researchers put forward the localization algorithm based on SVR.The data did not do any processing so that the positioning accuracy is not very high.In this paper,the other indoor positioning algorithm based on BP-SVR is presented,the data input SVR after inputting the BP neural network which playing an effect of data denoising and dimension transformation.Simulation experiments verified that the accuracy of the algorithm is 56.74%higher than that of the traditional LANDMARC algorithm,which is also 13.14%higher than that of the SVR-based indoor positioning algorithm.At the same time,the average positioning accuracy of the algorithm is superior to LANDMARC algorithm and SVR algorithm under different reference labels.3.On the basis of the concept of intelligent supermarket,this paper uses R2000 reader as the hardware platform,MySQL as the background database and Java as the programming language designing.It realizes and test the quick inventory,quick checkout and commodity positioning and also uses the similar data to achieve and test inventory forecasting and precision marketing.