Research And Application Of Key Technologies For Remote Maintenance Of Smart Vending Machine

With the advent of the network cloud computing era,"unmanned sales"is getting more and more attention.As a representative of unmanned retail terminals,In actual operation and maintenance in,there is a lack of terminal health monitoring,maintenance lag and inefficiency,arbitrariness large cargo commodity updates,high cost of operation and maintenance for a single smart vending machine and so on.This topic relies on the actual needs of enterprises,for above problems,this paper has studied 2 key technologies:smart vending machine operation process monitoring and product update intelligent recommendation maintenance,then,applied in the Hadoop distributed storage computing ecosystem,the research work completed in this paper mainly includes the following aspects:(1)For smart vending machine cabinet maintenance,Nagios monitoring system is designed to collect and store the terminal operation logs;Then,use a principal component modeling method to analysis the monitoring data,in the principal component space and the residual space extracting the SPE(Squared Prediction Error),T~2(Hotelling’s statistics),the mixed indicator as an index of evaluation indicators in the process monitoring,and use the control limits and variable contribution methods completion status monitoring and abnormal separation.(2)For smart vending machine recommendation decision maintenance,use the ALS(Alternating Least Squares)implicit feedback algorithms by matrix decomposition for an intelligent recommendation,analyze interest bias in the user evaluation matrix,we propose value affiliation model to measure the value of commodities to revise the user evaluation matrix trust,achieve the optimization of ALS recommended model.(3)Designed and built a big data processing Hadoop ecological computing framework,on this platform,use the terminal real data source to program and implement the monitoring and recommendation algorithm model.Implements terminal health monitoring and abnormal separation in terminal process monitoring;in the maintenance of product recommendation decision,offline optimized the ALS input parameters,the effectiveness of the ALS optimization recommendation model is verified by experiments,the experimental results show that the recommendation effect is better than artificial recommendation.(4)Designed and build ecosystem system of the smart vending machine remote maintenance,and developed Web application services based on this ecological architecture,achieve the following functions:maintenance tasks management,calculation data visualization,maintenance information mail synchronization,computing resource management and other functions,completed the application of key technologies of remote maintenance.In this paper,application of key technologies based on HDFS-Spark distributed storage computing framework,completed the implementation and deployment of core algorithms for key technologies,designed and developed the application services of smart vending machine remote maintenance,realized from the physical layer to the application layer ecological whole design and build,solved some problems in remote maintenance of samrt vending machine.