This paper proposes a maintenance platform for business vehicles which detects failure sign using IoT data on the move, orders to create repair parts by 3D printers and to deliver them to the destination. Recently, IoT and 3D printer technologies have been progressed and application cases to manufacturing and maintenance have been increased. Especially in air flight industry, various sensing data are collected during flight by IoT technologies and parts are created by 3D printers. And IoT platforms which improve development/operation of IoT applications also have been appeared. However, existing IoT platforms mainly targets to visualize "things" statuses by batch processing of collected sensing data, and 3 factors of real-time, automatic orders of repair parts and parts stock cost are insufficient to accelerate businesses. This paper targets maintenance of business vehicles such as airplane or high-speed bus. We propose a maintenance platform with real-time analysis, automatic orders of repair parts and minimum stock cost of parts. The proposed platform collects data via closed VPN, analyzes stream data and predicts failures in real-time by online machine learning framework Jubatus, coordinates ERP or SCM via in memory DB to order repair parts and also distributes repair parts data to 3D printers to create repair parts near the destination.

翻译：本文提出了一种面向商用车辆的维护平台，该平台利用行驶过程中的物联网数据检测故障征兆，通过3D打印机订购维修零件并配送至目的地。近年来，物联网与3D打印机技术取得显著进展，在制造业与维护领域的应用案例日益增多。尤其在航空产业中，飞行期间借助物联网技术采集各类传感数据，并通过3D打印机制造零件。同时，能够改善物联网应用开发与运行的物联网平台也应运而生。然而，现有物联网平台主要侧重于通过批量处理采集的传感数据可视化"物体"状态，在实时性、维修零件自动订购以及零件库存成本这三个方面存在不足，难以有效推动业务发展。本文以飞机、高速巴士等商用车辆的维护为研究对象，提出了一种集实时分析、维修零件自动订购与零件库存成本最小化于一体的维护平台。该平台通过封闭式VPN采集数据，利用在线机器学习框架Jubatus对流式数据进行实时分析与故障预测，通过内存数据库协调ERP或SCM系统完成维修零件订购，同时将维修零件数据传输至3D打印机，以便在目的地附近完成零件制造。