圖書典藏及影音數位平台

「牡丹水庫智慧營運與管理技術建置」計劃期程自106年至109年(共4年)，主要在整合水文觀測系統、監視系統、水質與水庫安全監測系統等加以改善，將水質及水量監控、水庫安全、防洪運轉等水庫營運相關資料以IOT(物聯網)方式，將資料傳輸至水庫管理單位雲端所建立之水庫資料中心，以利納入後續發展建置「多目標水庫智慧營運與管理」系統，創新並優化水庫營運管理，以供水庫安全與營運管理決策之用。 藉由改變現有傳統監測系統量測方式，配合物聯網新技術之整合與開發，達到監測與傳輸原件之微型化(MCU及MEMS感測器)及無線化， 監測與傳輸系統模組化，以達相關監測與營管數據雲端化，以利建構水質及水量監控、水庫安全及防洪運轉整合至智慧型營運管理平台系統，期藉由市場需求導向帶動相關產業技術整合與研發生產之發展，進而奠定擴拓國外智慧管理市場基礎。 第一期計畫(106年度)已初步在牡丹水庫建構水庫智慧營運與管理框架及智能化運作機制，第二期計畫(107年度)進一步執行感測器(含傳輸系統)等驗證及穩定性評估、建立水質及水庫安全營運(納入無人機智慧巡檢)預警指標，以及建立防洪運轉決策支援系統，第三期計畫(108年度)以達關鍵基礎設施維護、資安及水庫安全需求，建立水庫車輛監控及入侵安全辨識系統及系統資訊安全管理措施，並持續優化水庫智慧營運與管理架構，朝建構我國多目標水庫智慧營運與管理框架及智能化運作更完整之機制。

I. Origin and positioning of the plan The plan period of the “Mudan Reservoir Smart Operation-management Technology Construction and Establishment" was from 2017 to 2010 (a total of four years), primarily integrating and improving the hydrological observation system, monitoring system, water quality, and reservoir security monitoring system, and others. The data of the water dispatch, reservoir security, flood prevention and other information related to the reservoir operation were transmitted to the reservoir database on the cloud build by the reservoir management unit via IoT (Internet-of-Things). This would facilitate the “Multi-targets Reservoir Smart Operation-management” system that will be developed and constructed subsequently. It will innovate and optimize the reservoir operation-management, providing data for strategical uses of reservoir security and reservoir operation-management. By changing the existing measurement method of the traditional monitoring system and working in line with the integration and development of the new technology, IoT, it would achieve the miniaturization and wireless-ness of monitoring and transmission components (MCU and MEMS sensors), the modularization of the monitoring and transmission system and the cloud-based establishment of the relevant monitoring and operation-management of the data. This would facilitate the integration of the water dispatch, reservoir security and flood prevention operation into the smart operation-management platform system, hoping to set the foundation and expansion of the foreign smart management market the development of the technological integration and R&D production of the relevant markets led by the market demands. The preliminary plan (2017 and 2018) had initially established the smart operation-management framework and smart operation mechanism of the reservoir in Mudan Reservoir. This plan (2019) had further implemented the verification and stability assessments of the sensors (including the transmission system), established the water quality alarm indicators and reservoir security monitoring and control (including patent plate recognition, face recognition, and smart drone-inspections.), and build the flood prevention operation decision-making auxiliary system. This would optimize the smart operation-management structure of the reservoir, and build a multi-targets reservoir smart operation-management framework and a mechanism with a more comprehensive smart operation in Taiwan II. Water quality and quantity monitoring and control Currently, the water quality monitoring items of Mudan Reservoir include water quality parameters such as pH value, dissolved oxygen (DO), water temperature, chlorophyll-a, turbidity, blue-green algae fluorescence intensity, and electrical conductivity. Based on the analysis and results of the 2018-2019 monitoring data of the water floating platform, the data reliability of the water quality sondes, the correction suggestions, and the thresholds and warning values for each water quality item could be provided to the authority or the operations as a reference of contingency operations. Also, the debugging of the original data was recommended to be implemented this year. According to the current water quality monitoring data, except for DO which was quite high, the rest of the items were within the reasonable state; however, further observation of the water quality must be continued. This year, in addition to the newly established Mudan River turbidity monitoring station, analysis of the turbidity monitoring results of both Ru-reng River and Mudan River was also conducted. It showed that the current source of turbidity of Mudan Reservoir should be from Mudan River. The turbidity changes of Mudan River must be observed subsequently. III. Reservoir security (I) Verification and stability assessments of the perception layer and transmission system Periodic maintenance, component repair, and system optimization were conducted to the reservoir security related-equipment constructed on the first two terms in this year; statistical analysis results and slope warning values were also updated. According to the analytical results of the moisture monitoring platform No. 135 and of the statistics of the reservoir water level and precipitation, the cause of the moisture of platform No. 135 was due to precipitation rather than the water level of the water retention. This point further confirms the previous conclusion. Taking the model that takes into account the impact of downstream infiltration during the rainy season on the W2 measuring weir flow as a monitoring indicator of core integrity, the observation of limited strong rainfall during this year showed that the model is feasible. In 2019, the forecasting frameworks of the cumulative rainfall and groundwater level changes were updated for the forecasting model of water level changes caused by rainfall; the results showed that the rise of water level is related to the precipitation features, especially the cumulative delay of the infiltration rainfall intensity exceeding the threshold. Until now, the data of the plan is still limited. It is suggested to improve the accuracy based on the subsequent continuing monitoring results. The analytical results of the slope stability on the slopes of the right ridge and route C suggest the early warning value of slope stability based on the groundwater level. The results of the observation show that the shallow inclination of the right ridge reaches 10 degrees, belonging to a shallow sliding, which means only a slight impact towards the reservoir. Also, a simple cloud rainfall station was tested in the area of the reservoir this year; the results meet the requirements of simple installation and instant data transmission. (II) Important control facilities reaching to level two critical foundation safety protection In this plan, patent plate recognition, intrusion detection, face recognition, and image monitoring system were constructed inside Mudan Reservoir as important control facilities. Images can be exported via the photographic equipment to the image identification and analysis modules. This would manage and control the vehicle/personnel entry and exit at key regions. When the system identifies abnormalities (such as non-white list patent plate appearing at the reservoir region, people not in the permission list breaking into the management center, or other), the warning mechanism of the smart platform of this plan will be automatically connected in series, and the management unit will be notified via multiple warning methods to proceed to security protection actions. (III) Drone patrol system This year, an autonomous drone patrol system was established accompanied by an automatic charging function of the drone at the ramp. This will allow the drone to carry out a completely automatic flight and precise landing at the designated ramp for charging. Through the intuitive operation interface, the patrol mission of the drone can be set to reduce the risk of human operation negligence, achieving the goal of autonomous, safe, reliable realization human-machine IoT. Drone aerial image data of the reservoir and catchment area have been collected in this term to be used as the training data of future AI smart reading. IV. Flood prevention operation (I) The smart positioning of flood control operations consists of "Comprehensive collection of useful information to be provided to the managers for a correct situational judgment as well as the planning of a more efficient, effective and thoughtful reservoir water discharge decision-making." The necessary information includes primarily precipitation (rainfall station or radar), reservoir water level, inlet flow, gate opening, and discharge flow, downstream river level, etc. The details covered by the decision-making analysis include observation error detection and exclusion, occurred rainfall assessment, future rainfall forecasting in the reservoir catchment area, simulation of rainfall discharge in the reservoir catchment area, and assessment of reservoir flood prevention operation strategies. The output information can propose future water discharge strategies and suggestions that can take into account both reservoir security and reduction of downstream floods, reaching the goal of water retention at the final term as well as the maximization of the effectiveness of sand and sludge removal with the discharge. (II) The research and development of the decision-making analysis in this year was focused on the possible error detection and exclusion of the observed precipitation during the precipitation calculation during floods. Such method adopted the concept of inverse problem. Based on the observed inflow of reservoir and rainfall-runoff pattern in the catchment area, the unsupervised and supervised methods were used separately with the aim to effectively increase the precision of the simulated flow amount. This would automatically determine the timing in which error might have occurred at the precipitation station. An automatic interface with relevant rainfall stations and the hourly reservoir water level data in a web form was developed, and the future rainfall-runoff procedure estimation and the average precipitation at the catchment area could be automatically analyzed and presented; this could be provided as a reference of flood prevention operation. The developed method was adopted to assess and analyze the 35 historical cases. The simulation efficiency coefficient was increased to 0.91 on average in every case. The average runoff coefficient was reduced to 0.70, which represented a significant improvement over the current situation. (III) To provide instant automatic analysis of the results of precipitation, this plan had completed the CWB disaster prevention radar data interface and display website development at Liyuan. By comparing the precipitation observation at the catchment area of Mudan Reservoir during the Severe Tropical Storm Bailu and the precipitation observed with the disaster prevention radar, the difference of the accumulated precipitation of both was 115 ~ 235 mm, which means clear biased estimate; the correlation coefficient is 0.63 ~ 0.90, which means high correlation, showing the high possibility to detect rain trend and peak rainfall time with the disaster prevention radar; however, there is still underestimation in terms of peak precipitation amount and the accumulated precipitation. In regard to the Mudan Reservoir combined forecasting rainfall data interface analysis and display website development, there are three types of combined precipitation forecasts: QPESUMS_QPF, QPESUMS_WRF, and QPESUMS_ETQPF, providing long-term precipitation forecast of the Mudan Reservoir catchment area. QPESUMS_QPF consists of the comparative analysis and display of the error between the observed and the forecasted precipitation. The comparative analysis of the observed and the forecasted precipitation with QPESUMS in the case of tropical storm Bailu in Mudan Reservoir, there was an overestimation of the accumulated precipitation at every hour in the hourly future forecasting. In particular, the difference of the accumulated precipitation at the second hour was the largest (93.6 mm). It was the same for the average precipitation difference, having the largest difference (8.2mm) at the second hour in the hourly future forecasting. The difference of the first and the third hour of the hourly future forecast was approximately 7.0mm. The relevant coefficient reduced along with the time of the forecasting, having the highest (0.6) at the first hour of the hourly future forecasting and the lowest (0.4) at the third hour. It is visible that the best time to understand the trend was the first hour of the hourly future forecast. In regard to the display website development of the comparative results of the instant precipitation and rain-flow of Mudan Reservoir, it provides the flood prevention staff to be updated with the precipitation forecast of Mudan Reservoir in terms of the combination of reasonable estimates and best (reasonable) rainfall stations of every rainfall stations. In addition, it also interfaces with observational water information in the downstream of Mudan Reservoir (Sichongxi Basin), and the instant automatic flow forecast was constructed and developed together with the simulation results of the Mudan decision-making simulation system. The calculation output file (output.csv) of the Mudan Reservoir operation strategy is uploaded to the FTP space at every hour fix, providing the bureau and other flood prevention units at the downstream a reference and application. V. Expansion, improvement, and construction of the reservoir smart operation-management system The reservoir smart operation-management platform has been integrated with the reservoir electrical facilities, peripheral sensors, precipitation stations of the Central Weather Bureau, the water level stations of the Water Resources Agency (WRA), reservoir data (storage capacity, gauge and water rights), and others for further analysis and application that would be provided as suggestions to the managers to adjust the water dispatch, reservoir security, flood prevention operation decision-making, achieving the goal of smart operation-management. The plan at this term is to continue implementing the optimization and improvement of the smart management platform interface. Through the creation of a platform homepage real-time dashboard with image and web component design, it can visualize and display the observed values of the reservoir and update them with highly-precise diagrams with graphic webpages. This would achieve a consistent presentation that could transmit instant observed information with precision. Regarding the warning interface integration planning, different warning conditions, contents, and recipient-groups can be set based on different warning settings. Also, the recipient modules of the defined events are integrated with warning (earthquake trigger, face recognition, patent plate recognition, humanoid intrusion detection) of the subsystem so that the smart management platform can deliver the warning uniformly. By integrating the security planning of the key infrastructure of the reservoir in this term, the platform has included the security camera live images, historical images, and patent plate identification record query, avoiding human-damages to the facilities and ensuring stable water supply. Regarding the wireless transmission facility improvement, further assessment was carried out in this term to assess and draft the improvement measures for the existing wireless communication and transmission; equipment data update module was developed so as to upload the data of the sensors directly to the WRA. Also, wireless communication facility connection management service was established to be updated with the status of the wireless communication. In line with WRA’s “Water Resources IOT Sensor Infrastructure Cloud Operation Network” plan, the team has also applied cloud web hosting VM resources to the National Center for High-performance Computing (NCHC), completed the application settings of the operation systems, anti-virus software installation, and firewall, and completed the establishment of the smart management platform website and data transfer, making it possible to continue uploading the data of the sensors to the IoT platform. VI. Cybersecurity level meets Class B requirements In this term, cybersecurity safety inspection, health check, threat detection, and management mechanism establishment, the introduction of government configuration benchmarks, the establishment of cybersecurity protection, cybersecurity education and training, and deployment of security credentials were implemented: (I) Cyber security safety inspection included infiltration testing of four systems, weakness screening of five websites, and weakness screening of fourteen hosts, ensuring that there were no high-risk weaknesses. Repairs were carried out to the systems and hosts with high risks detected at the preliminary testing. Afterward, a secondary testing was conducted to make sure the safety of the fortified systems and hosts. (II) In the cybersecurity health-check, one overall inspection was conducted on the network structure, wired network malicious activities, user-end computers, server hosts and safety setting of this plan, finding out the hidden problems and providing improvement suggestions. (III) In the threat detection and management mechanism establishment and the establishment of cybersecurity protection, the next-generation firewall at network exit provides intrusion detection and protection mechanism, reducing the risks of external hacker intrusions to the internal systems and hosts. (IV) In the introduction of government configuration benchmarks, one set of government benchmark management system was established, and the policy was applied to the host in order to ensure the consistency of the security setting at all terminals and to be in line with the cybersecurity policy of the Executive Yuan. (V) Cybersecurity education and training were provided with three-hour cybersecurity latest news home learning. The contents covered cybersecurity case studies, trends, blackmail software, abduction mining, mobile security, IoT security, personal computer security improvement, and other issues. This would raise the personnel’s cybersecurity consciousness and alert and reduce the possibility of cybersecurity incidents due to unintentionally triggering of malicious e-mails or websites. (VI) In terms of the deployment of security credentials, three TLS/SSL certificates were applied and deployed at smart management platform VM, flood-prevention operation subsystem, and drone subsystem, encrypting the communication between the users and the systems and avoiding the interception and eavesdropping by third parties. The aforementioned mechanisms covered the security at the network, terminal, development, transmission, and personnel cybersecurity levels, which could enhance cybersecurity of this plan and meet the regulatory goal of Class B cybersecurity responsibility of the public authorities.