圖書典藏及影音數位平台

移動式抽水機是積淹水清除的重要應變設備，每當颱風豪雨發生時，為因應各地淹水情勢，移動式抽水機必須互相支援，靈活調度以求設備效能有效發揮。因此應用發展成熟之全球衛星定位系統、無線通訊與地理資訊系統等現代科技，以提昇移動式抽水機管理。規劃建構「移動式抽水管理系統計畫」，以即時提供抽水機動向及勤務工作狀態紀錄資料，相關紀錄並可作為日後路線規劃及輔助決策之參考，監控中心執勤人員可於即時通報特定單位相關訊息，俾利掌握搶救先機。 本年度辦理本署暨代辦臺北市及桃園市政府大型移動式抽水機裝設車載機監控模組，可即時提供抽水機動向及勤務工作狀態紀錄資料，相關紀錄並可作為日後路線規劃及輔助決策之參考，監控中心執勤人員可即時通報特定單位相關訊息，俾利掌握搶救先機。

The water pumping machines were known as an important and efficient flooding control equipment in Taiwan. When flood occurred, the River Bureau and local governments have to coordinate the water pumping machines to eliminate floods. For this reason, Water Resource Agency utilized the GPS (Global Positioning System), wireless communication and GIS (Geography Information System) technologies to execute the project. The project which started form April 4th, 2018 to December 25th, aimed to equip with GPS device on water pumping machine and transferred data via 4G networking to cloud system. Disaster response operators could track the locations, operation records of machine and get the instant reports from web or App. The whole project required 84 GPS devices to install on water pumping machines, including 65 for all River Bureaus, and 9 for Taipei City Government, 10 for Taoyuan City and cloud system for tracking and statistics during the period of emergency response. This project had finished 84 GPS devices installation and construction, Water Resources Agency water pumping machine management system platform and its extended features before May. The project completed GPS devices scheduled maintenance from Aug 6th to Aug 13th and Oct 15th to Oct 23rd and executed regular maintenances report. And held 3 training sessions (2 external trainings and 1 internal training) to enhance operator’s skills for devices and system. And also executed regular maintenances on August and October, and delivered maintenance reports accordingly. Also this project required assistant personnel during emergency response periods. There were 7 emergency response assistances and 426 man-hours in total and aim at 0823 torrential rain information compilation. In addition to those, this project also proceeded to analyze historical data for machine learning process by collecting data set of machines’ locations and rainfall forecast raster data from QPESUMS. Then design the statistical model and algorithm to find out the relevance of rainfall and machine dispatch. Furthermore, this project also upgraded the architecture of cloud software by adopting open sources data analytics cluster Spark and No-SQL Cassandra to enhance the performance of the data processing and web browsing. According to the experiences of 0823 flood accident emergency response this year, this project suggested the following items: 1. draw up a detail plan for the water pumping machine operation SOP (Standard Operation Procedure) in order to provide guidelines for the machine operation crew; 2. Develop the web features for local governments to provide dispatch information while the severe incidents strike. 3. Estimate the possibility for water pumping machine by IoT (Internet of Things) sensor and LPWAN (Low-Power Wide-Area Network) adoption for the future. This project also suggests extending the study of historical data analysis for better artificial intelligence application.