TWI284865B - Real-time mobile debris-flow disaster prevention and alert system - Google Patents

Real-time mobile debris-flow disaster prevention and alert system Download PDF

Info

Publication number
TWI284865B
TWI284865B TW94104363A TW94104363A TWI284865B TW I284865 B TWI284865 B TW I284865B TW 94104363 A TW94104363 A TW 94104363A TW 94104363 A TW94104363 A TW 94104363A TW I284865 B TWI284865 B TW I284865B
Authority
TW
Taiwan
Prior art keywords
information
agent
earth
user
multimedia
Prior art date
Application number
TW94104363A
Other languages
Chinese (zh)
Other versions
TW200629183A (en
Inventor
Hsu-Yang Kung
Hau-Shiang Gu
Jing-You Lin
Guang-Rung Tsai
Original Assignee
Nat Pingtung University Of Sci
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nat Pingtung University Of Sci filed Critical Nat Pingtung University Of Sci
Priority to TW94104363A priority Critical patent/TWI284865B/en
Publication of TW200629183A publication Critical patent/TW200629183A/en
Application granted granted Critical
Publication of TWI284865B publication Critical patent/TWI284865B/en

Links

Abstract

This invention discloses a real-time mobile debris-flow disaster prevention and alert (RMD2PA) system. The RMD2 PA system is a 3-tier structure composed of a mobile user terminal system, a multimedia application server, and an expert decision system. A user at the mobile user terminal system can use a portable device to bi-directionally transmit the multimedia information of the debris-flow at a disaster area via GSM/GPRS personal mobile communication network. The system provides several functions such as an AI proxy provided by a multimedia application server used to pre-process and examine data to save the bandwidth of a wireless network so as to fulfill the customization and personalization requirement, an analysis engine used to simulate a debris-flow path via coordinates and the comparison of images, visual reality used to provide relative landform simulation service facilitating an Internet user to connect RMD2PA system to access relative debris-flow information, and a Web and WAP service using Web and WAP functions to enter an interactive web-page to access relative information, so as to sort mass multimedia information to save the time used to access data in a post-terminal database. Moreover, an expert decision strategy server at the post-terminal Internet is introduced to numerically analyze and construct the simulated model and the mechanism of debris-flow. The system can use the information techniques of a global positioning system (GPS), a geographic information system (GIS), and a remote sensing (RS) to determine the value of the parameter of debris-flow disaster factors that are calculated by regression analysis, so as to achieve precise disaster prevention and salvage guiding functions.

Description

-^84865 九、發明說明: 【發明所屬之技術領域】 本發明係-種仃統土石流災情丽與魏线,其尤指整合多媒體 p時㈣地理貝錢統、衛星驗系統、遙感探測、無線分封協定、智 慧型代理、虛擬實境、逃生路徑與推理引料,並應用於土石流災害之預 方”通報使火區之火害能即時有效之傳遞,進行辅助災民撤離危險之區 域,搶救人民生命財產安全。 【先前技術】 叙的自然X害中’其發生包含了地質因素者,通稱為「土砂災害」, i括地表土壤冲餘、崩塌'地滑、土石流、河川淘刷與堆積等。其發生 之讀與規模,除了與降雨強度及雨量直接相關外,陡λ肖之坡地及地質脆 弱地區尤紐生。另外,人為之開發糊,諸如··伐木、墾植、開路、採 礦,及其他有關開發行為,加上不良的水土保持設施,則極易誘發坡地的 土砂災害。臺灣的坡地,尤其以中高海拔地區,雨量充沛、地形陡惰且地 質脆弱,因此崩塌、地滑等災害頻仍。其若為地表之均夷侧,應屬自然 見象’然其母母因人為之不當開發或水土保持設施不良,誘使災害之發生, 災情加劇。 土石流災害之發生,乃因泥、砂、礫及巨石等地質材料與水之混合物 又重力作用後產生流動所造成之災害。其發生包括了三個要件·⑴充足 的水里’⑵足夠的土方;(3)有效的河床坡度。充足的水量通常源自 降雨強度與累積雨量’而有效的河床坡度姻地而異;臺灣地區之土石流, .1284865 、可床皮度1G度、集水面積1Q公頃以上地區較綠發生。而集水面積即 …月著水里的大小與流速的快慢;其巾,足觸土謂指的乃是河流上游 /可奋中堆積物的量,而河紅游之堆積物來齡了絲土壤賴所殘留於 /可谷中者外’其最主要之來源乃為河流上游邊坡土石關塌或地滑而堆積 於〉可谷中者。崩戦地滑乃指河谷邊坡之岩層因重力加上氣翻素(如: 暴雨),而崩落或滑落之現象。 土石流之發生,除了降雨的因素外,並與地質環境有相當的關係。以 臺灣處於非常不安梅也質環境中,山高水急,加上各項地質活動頻仍, 使得自然的地質災害,如:崩塌、落石等習以常見。因此,土石流災害發 生所需之土方不缺,此亦為臺灣土石流災害頻繁的原因之一。 而其近幾年來臺灣土石流災害曰益嚴重,往往一場豪雨即引發重大土 石机火害ie成人民生命財產的嚴重損失,為此相關單位投注大筆經費建立 石机火障預防與通報系統,以當土石流災害發生時,可達到預先馨告而 減輕人民生命財產損失的目標。1G年前設置的18座土石流監測系統,由於 故障率高且準確率未達30%,所以目前只能作為研究上的參考值,然而當 真正發生時,土石流災區的通訊線路往往斷線,無法有效進行災情 通報與警示。 【發明内容】 《所欲解決之技術問題》 此外由於台灣人口聚集稠密,災區已不限於偏遠區域,例如gg年的象 神趟風對汐止山頂大片住宅區造成水災與土石流重大災情,而今年的敏督 -1284865 利亦同,因此在某些地區中,例如人口稠密之山坡地住宅區與已有部份通 訊基礎建設之山區(例如南投縣大部份山區),如何利用最簡便的通訊與資 訊設備(例如PDA加手機),實現有效的即時(Real-Time)災情多媒體資訊(包 括為料、聲音與照片)傳輸以及災情決策與預防,已經是國科會國家型防救 災計劃所擇定的重要議題之一。因此當土石流真正發生時,會面臨下列的 情況:(1)災區訊息通訊不易且混亂,喪失資訊傳遞功能,以致災情未能 直接即時通報至各防救災中心,而延誤救災時機,故達不到系統的成效, 造成災區嚴重的損傷。(ii)網際網路建置的決策支援系統,或預警系統需 要桌上型電腦或手提式電腦利用有線網路來做連結與輸入資料,一但有線 網路發生問題或操作不當,就無法使用。而且在災區蒐集現場資料,需在 豕中或特定地方的電腦做輸入,無法達到即時性的輸入資料,造成警訊並 不能即時通知。(iii)使用昂貴之衛星電話進行救災訊息傳送,也往往由 於使用者的操作疏忽以及衛星電話有限功能之限制,而無法發揮預期之功 月b。例如位於災區之衛星電話使用者不小心將衛星電話關機,災情預防與 通報中l即無法預先主動通知示警,而昂貴的衛星電話通訊品質往往不良 (尤其是災害發生天候惡劣時),此外衛星電話僅提供通話功能,因此無法 提供災情預防與通報中心所需要之災區現場的第一手數據資料。(iv)說明 有效的土石流決策系統需有即時之現場資訊,並採決策評估與推理機制以 正確地研判災情並作出迅速而有效的災情預防措施。(v)受災之民眾不知 該往何處進行疏散,且災民逃生方向不定,使得救災更佳困難,因此必須 提供災區人民如何迅速地取得逃生訊息指示,例如適當逃生路徑等資訊。 1284865 因此尚未有一套可以有效整合無線行動通訊及災情資料庫之土石流決策資 汛系統,因此如何建立有效不受時空限制,而且具即時性之土石流災害預 防通報系統將是最大的課題。 《對於先前技術之效果》 有鑑於上述之發明背景中之其它土石流災害預防與通報系統之缺點, 本發明係由行動式使財端系統、多媒體應用铜服器與專家決策系統所組成, 其中行動式使用者系統的使用者利用手持設備,可以經由GSM/GPRS個人 行動通X網路’進行災區雙够媒體土石流資訊傳收,在透過—多媒體應用伺 服為内所提供的智慧型代理(進行資料先前處理與過滤,以節省無線網路頻寬並 達成客製化與個人化需求)、推理引擎(經由座標與圖像比對方式推測土石流行 進路徑)、虛擬實境(相關地形模擬服務,讓網際網路的使用者連結1^1021)八系 統存取土;5她職訊)、Web化服務與WAP服務(糊Web及WAp的功能進 入互動式網頁存取相關資料)等功能,達到分離巨量之多媒體資訊以節省後端 之資料庫存取資料之時間,祕合網際網路後端的—專家決策伺服器利用 數值分析法建立土石流推論模式與機制,藉由全球衛星定位系統(Gps)、地 理資訊系統(GIS)與遙感探測(RS)資訊技術找出土石流災害因子參數值進行 回歸分析與驗㈣發法,以達到精麵災害制與救災指引功能,利用 上述系統提出之發明除解決習知之問題外亦提供更即時與更準確之服務, 於設計時即考慮高度可擴充性與行動式多舰等侧功能。本發明之目的 為整合多舰㈣影音、地理資訊系統、衛星定㈣統、遙祕測、無線 分封協定、智慧型代理、虛擬實境、逃生路徑與推理引擎等,並應用於土 1284865 -石流災害之預防與通報,使災區之災害能即時有效之傳遞,進行辅助災民 -撤離危險之區域,搶救人民生命財產安全。 【實施方式】 如第-圖並配合第二圖所示,一種行動化土石流災情預防與通報系統 共採三層次架構(3-tier) ’第一層為行動者使用系統1〇,其使用者可以利 用各種終端設備進行存取遠端的防災⑽資料,其中包括智慧型手機 籲(Smart Phone)、個人電腦(Pe聰al c寧他,pc)、筆記型電腦⑽㈣、 平板電腦(Tablet PC)與個人數位助理(p⑽如叫制⑻伽的,舰) 等存取第二層多媒體應用舰器2〇與專家決策系統3〇中相關資料。多媒 體應用伺服器20中有提供智慧資訊代理、多媒體虛擬實境與齡舰器, 專家決策祕财擁有智慧推論引擎輯土^流災害發生絲及其它危險 度之推論’如此之設計是讓系統更具實用性與有效性,以下進行述說每一 部份之設計。 一、行動式使用者端系統設計 11Π如ί二圖/斤,I行動式使用者端所提供之功能包括基礎座標服務 rViCe)、客製化服務 12(Customized Service)、跨 &網路13、即時多媒體傳輸η與推論引料,以下分別論述其詳細)相關 座標服務··行喊賴者湘手邊的終端賴連結至行動通賴 網際網路,登入並使用本系統,並藉由GPS自動進行所在地區定位。 f /^寻巧之座標則利用GSM/GPRS手機經大哥大行動通訊網路傳回到後 知專豕決策資料庫中,進行地區資訊分析。此功能在儲存每一位使用者 位置,長:供基礎座標服務,用以了解每一位使用者所在地區盘所在地資 訊,再經由後端資料庫進行決策分析後匯整出一張表單,將g有線使 用者列出,並可知道該使用者是否位於危險區域中。 1284865 (2) 客製化服務:當系統擁有使用者所在地資訊同時提供客製化 (Customized)相關服務,包括依使用者端設備不同提供不同之畫面、所 位於地區是否有土石流災害與新聞下載等資訊,讓使用者不需經繁瑣的 ‘ 步驟將相關資訊擷取下來。 (3) 跨異質網路:跨異質網路進行資料傳輪,因為使用者所處之網路不同, 因此系統之設計讓終端設備或行動設備可由全球通訊網路(Global-^84865 IX. INSTRUCTIONS: [Technical field of invention] The present invention is a kind of earth-rock disaster and the Wei line, especially when integrating multimedia p (4) geography, satellite system, remote sensing, wireless Separate agreements, intelligent agents, virtual reality, escape routes and reasoning materials, and applied to the pre-planning of earth-rock disasters, the fires can be transmitted instantly and effectively, assisting the victims to evacuate dangerous areas and save people's lives. Property security. [Prior Art] Among the natural X harms mentioned, 'the occurrence of geological factors, commonly known as "soil disasters", including surface soil flushing, collapse 'slip, earth and rock flow, river wash and accumulation. The occurrence and scale of its occurrence are not only related to rainfall intensity and rainfall, but also to the slopes of the steep λ and the geologically fragile areas. In addition, artificial development of pastes, such as logging, planting, road opening, mining, and other related development activities, coupled with poor soil and water conservation facilities, can easily induce soil sand disasters on slopes. Taiwan's sloping land, especially at medium and high altitudes, has abundant rainfall, steep terrain and fragile geology. Therefore, disasters such as collapse and ground slip are frequent. If it is the uniform side of the earth's surface, it should be a natural appearance. However, its mother and mother are improperly developed due to human beings or poor soil and water conservation facilities, which induces disasters and intensifies the disaster. The occurrence of earth-rock disasters is caused by the mixture of geological materials such as mud, sand, gravel and boulders and water, and the flow caused by gravity. Its occurrence includes three elements: (1) sufficient water in the '(2) sufficient earthwork; and (3) effective riverbed slope. Sufficient water is usually derived from rainfall intensity and accumulated rainfall, and the effective riverbed slope varies from place to place; the earth-rock flow in Taiwan, .1284865, the bed thickness of 1G, and the catchment area of 1Q hectare are more green. The catchment area is the speed and flow rate of the water in the moon; the towel, the foot touches the soil, refers to the amount of sediment in the upper reaches of the river, and the accumulation of river red swims is the age of silk. The main source of the soil remains in the valley of the river. The main source of the soil is the collapse of the upstream slope of the river or the slippage of the ground and the accumulation in the valley. Collapse and slippage refers to the phenomenon that the rock layer of the river valley slope collapses or falls due to gravity plus gas-turning (such as: heavy rain). The occurrence of earth and rock flow, in addition to the factors of rainfall, has a considerable relationship with the geological environment. In Taiwan, which is in a very uneasy environment, high mountains and high waters, coupled with frequent geological activities, make natural geological disasters such as collapse and falling rocks common. Therefore, there is no shortage of earthwork required for earth-rock flow disasters, which is one of the reasons for the frequent earth-rock disasters in Taiwan. In recent years, the disasters of earth-rock disasters in Taiwan have been very serious. Often, a heavy rain has caused serious damage to people’s lives and property, and the relevant units have invested heavily in establishing a stone fire prevention and notification system. When the earth-rock flow disaster occurs, the goal of reducing the loss of life and property of the people can be achieved by the advance notice. The 18 earth-rock flow monitoring systems set up before 1G, because of the high failure rate and the accuracy rate of less than 30%, can only be used as a reference value for research. However, when it happens, the communication lines in the earth-rock disaster area are often broken. Effectively report and alert the disaster. [Summary of the Invention] "Technical Issues to be Solved" In addition, due to the dense population of Taiwan, the disaster area is not limited to remote areas. For example, the hunter-like hurricane of gg has caused major disasters such as floods and landslides in the large residential areas on the top of the mountain. Minto-1284865 is also the same, so in some areas, such as densely populated hillside residential areas and some mountain areas with some communication infrastructure (such as most of the mountainous areas in Nantou County), how to use the simplest communication and Information equipment (such as PDA plus mobile phone), to achieve effective real-time (Real-Time) disaster multimedia information (including materials, sound and photos) transmission and disaster decision-making and prevention, has been selected by the National Science Council National Disaster Prevention Plan One of the important topics. Therefore, when the earth-rock flow really occurs, the following situations will be encountered: (1) The information communication in the disaster area is not easy and chaotic, and the information transmission function is lost. As a result, the disaster situation cannot be immediately reported to the disaster prevention centers, and the disaster relief time is delayed. The effectiveness of the system has caused serious damage in the disaster area. (ii) Internet-based decision support systems, or early warning systems that require a desktop or laptop computer to use a wired network to make connections and input data. Once the wired network has problems or is not operating properly, it cannot be used. . Moreover, in the disaster area, the site information is collected, and the computer needs to be input in the computer or in a specific place. The input data cannot be instantaneously generated, and the warning is not immediately notified. (iii) The use of expensive satellite phones for the transmission of disaster relief messages is often due to user negligence and limited functionality of satellite phones, and is not able to perform the expected function. For example, satellite phone users in the disaster area accidentally turn off the satellite phone, and the disaster prevention and notification can not promptly notify the police in advance, and the quality of expensive satellite phone communication is often poor (especially when the disaster occurs in bad weather), in addition to satellite phones. The call function is only available, so it is not possible to provide first-hand data on the disaster area required by the Disaster Prevention and Notification Center. (iv) Explain that an effective earth-rock flow decision-making system requires immediate on-site information, and decision-making assessment and reasoning mechanisms are used to properly determine the disaster situation and to make prompt and effective disaster prevention measures. (v) The affected people do not know where to evacuate, and the disaster-stricken people's escape direction makes the disaster more difficult. Therefore, it is necessary to provide information on how quickly people in the disaster area can obtain escape information, such as appropriate escape routes. 1284865 Therefore, there is no set of earth-rock flow decision-making information system that can effectively integrate the wireless mobile communication and disaster database. Therefore, how to establish an effective earth-rock disaster prevention and notification system will be the biggest issue. "Effects on Prior Art" In view of the shortcomings of other earth-rock disaster prevention and notification systems in the above-mentioned background of the invention, the present invention is composed of an action-based financial system, a multimedia application copper server and an expert decision-making system, wherein the action The user of the user system can use the handheld device to carry out the information transmission of the double-media media in the disaster area via the GSM/GPRS personal mobile X network, and provide the intelligent agent through the multimedia application servo. Previous processing and filtering to save wireless network bandwidth and achieve customization and personalization), inference engine (inferred geophysical trend path via coordinate and image comparison), virtual reality (related terrain simulation service, let Internet users connect 1^1021) eight systems to access the soil; 5 her job news), Web services and WAP services (paste Web and WAp functions into interactive web access related information) and other functions, to achieve separation A huge amount of multimedia information to save time in the back-end data inventory, the number of experts in the Internet back-end - expert decision server utilization The analysis method establishes the inferred model and mechanism of the earth-rock flow, and uses the global satellite positioning system (Gps), geographic information system (GIS) and remote sensing (RS) information technology to find out the parameters of the earth-rock disaster factor for regression analysis and verification (4). In order to achieve the functions of the Meteorological Disaster Relief System and the Disaster Relief Guidelines, the inventions proposed by the above system provide more immediate and more accurate services in addition to solving the problems of the prior art, and are designed with high scalability and mobile multi-ship functions. The purpose of the invention is to integrate multi-ship (four) audio and video, geographic information system, satellite (four) system, remote secret measurement, wireless packetization protocol, intelligent agent, virtual reality, escape path and inference engine, etc., and applied to soil 1284865 - stone The prevention and notification of disasters will enable disasters in the disaster areas to be transmitted immediately and effectively, and assist the victims to evacuate dangerous areas and rescue people's lives and property. [Embodiment] As shown in the figure - in conjunction with the second figure, a three-level structure (3-tier) is adopted for the action-based earth-rock disaster prevention and notification system. The first layer is the actor's use system. You can use a variety of terminal devices to access remote disaster prevention (10) data, including Smart Phone, PC (Pe Cong, C), Notebook (10) (4), Tablet PC (Tablet PC) And the personal digital assistant (p (10) such as the system (8) gamma, ship) access to the second layer of multimedia application ship 2 〇 and the expert decision system 3 相关 related information. The multimedia application server 20 provides a smart information agent, a multimedia virtual reality and an age ship. The expert decision-making secret wealth has a smart inference engine, and the inference of the disaster occurrence silk and other dangers is so designed to make the system more Practical and effective, the following describes each part of the design. 1. The design of the mobile user system is as follows: the functions provided by the I mobile user include the basic coordinate service rViCe), the customized service 12 (Customized Service), the cross & network 13 , instant multimedia transmission η and inference primers, the following respectively discuss the details of the relevant coordinates service · · screaming the stalker's terminal to connect to the mobile Internet, log in and use the system, and automatically through GPS Location in the region. The coordinates of f / ^ 巧巧 use GSM / GPRS mobile phones to pass back to the post-decision decision-making database through the Big Brother mobile communication network for regional information analysis. This function stores each user's location, length: for the basic coordinate service, to understand the location information of each user's local area, and then through the back-end database for decision analysis, a form will be collected, will g The cable user lists and knows if the user is in a hazardous area. 1284865 (2) Customized service: When the system has the user's location information and provides customizedized services, including different screens depending on the user equipment, whether there are earth-rock disasters and news downloads in the area. Information, so that users do not have to go through the cumbersome 'steps to get the relevant information. (3) Cross-heterogeneous networks: data transmission across heterogeneous networks. Because users are in different networks, the system is designed to allow end devices or mobile devices to be connected to the global communication network.

System for Mobi le Communications, GSM)、無線分封交換服務(General Packet Radio Service, GPRS)、有線網路、IEEE802. llx 無線網路等 環境進行整合使用。 (4) 即時多媒體傳輸:即時多媒體傳輸部份包含文字(Text)、圖片(Picture) 舆影音(Video and Audio)三部份進行資料的傳輸,使用者可以利用一 般文字進行線上交談並將訊息群播(Multicast)傳送給位於同一區域的 群組所有成員(Group Members)。此外亦可利用PDA上的照像機將災區 φ 第一現場現地情況拍攝下來,即時地傳送回後端予專業人員進行研判, 或利用影音傳送之功能將現場影音傳送回後端專家決策系統與專業人 員,以利相關第一手災情之研判。 (5) 推論引導:若行動通訊網路斷線或無法連接後端專家決策系統時,手 持設備上亦提供由資料探勘所擷取出之土石流災區危險因子,架構出之 推理機制進行現況之推估。 二、多媒體應用伺服器之設計 如第四圖所示,位於網際網路前端之多媒體應用伺服器2〇主要提供多 媒體相關服務,其包含智慧型代理21、虛擬實境22、Web化服務23、WAp 服務24與推理引擎等功能。其目的在分離巨量之多媒體資訊以節省後端之 資料庫存取資料之時間。 (一)智慧型代理(Intelligent Agent) $慧型代理進行資料的絲、搜尋、分類、處理或通報等工作,能迅 速確實地讓使用者掌握最即時資訊。如第五圖所示,本祕所設計之智慧 代21匕括⑴使用者介面代理批⑽奸Agent· UIA)、⑵ 1284865 ' 監控代理 213(Monitor Agent)、(3)雨量資訊代理 214(Rainfall Agent)、 • (4)新聞資訊代理215(News Agent)、(5)緊急通報代理2i6(AlertSystem for Mobi le Communications, GSM), General Packet Radio Service (GPRS), wired network, IEEE802. llx wireless network and other environments for integration. (4) Instant multimedia transmission: The instant multimedia transmission part includes text, picture and video and audio. The user can use the general text to conduct online conversation and the message group. Multicast is delivered to all members of the group in the same area. In addition, the camera on the PDA can be used to capture the first scene of the disaster area φ, and then immediately transmitted back to the back end for professional research, or use the function of video transmission to transmit the live video to the back-end expert decision system. Professionals to facilitate the research and judgment of the first-hand disaster. (5) Inference guidance: If the mobile communication network is disconnected or unable to connect to the back-end expert decision-making system, the risk factors of the earth-rock disaster area extracted by the data exploration institute are also provided on the handheld equipment, and the reasoning mechanism is constructed to estimate the current situation. Second, the multimedia application server design As shown in the fourth figure, the multimedia application server located at the front end of the Internet mainly provides multimedia related services, including a smart agent 21, a virtual reality 22, a Web service 23, WAp service 24 and inference engine and other functions. Its purpose is to separate huge amounts of multimedia information to save time in the back-end data inventory. (1) Intelligent Agent (HP) Agents can perform the task of silk, search, classification, processing or notification of data, and quickly and reliably let users grasp the most instant information. As shown in the fifth figure, the smart code designed by the secret agent 21 includes (1) user interface agent batch (10) agent Agent · UIA), (2) 1284865 'monitoring agent 213 (Monitor Agent), (3) rainfall information agent 214 (Rainfall) Agent), • (4) News Agent 215 (News Agent), (5) Emergency Notification Agent 2i6 (Alert

Agent)、(6)預先下載代理 217(Pre-l〇ad Agent)、(7)協調代理 218(Collaboration Agent,CA)與(8)圖層資訊代理 219(PlateAgent), (6) pre-download agent 217 (Pre-l〇ad Agent), (7) Coordination Agent 218 (Collaboration Agent, CA) and (8) Layer Information Agent 219 (Plate

Information Agent,PIA)等八個,其工作與運作流程如下述說之·· (1)使用者介面代理:分辨用戶端是使用那一類型設備,當用戶端連上線 時’便會給予適當的操作界面與内容; _⑵監控代理:監控每-侧戶端的齡,若有人新連結上細通報使用 者介面代理,給予客製化(Cust〇mized)的使用資訊; (3)雨篁資訊代理:監控雨量站的資訊並隨時提供給各個使用者; ⑷新聞資訊代理:於網際網路之新聞資訊(例如γΑ_奇摩—新聞)自動抓 取土石流相關㈣,制戶端織看最近所發生的緊急事件; ⑸緊急通報代理:若某一區推論發生土石流事件,則會回傳給位於該縣 市區域所有或特定用戶端,並通知相關數據資料與情形; ⑹預先下載代理·使用者可以依所位於地區來下載雨量或新聞相關資 料’若有更新則由代理自動傳遞新資料。 ⑺制代理·協贿用魏之傳收,避免相_資訊於㈤服器與用戶端 重覆存取; ()SU貝Λ代理·辅助進行圖層資訊下載,並依其設備不同給予不同之 圖層建4’、套遠’包括等高線地形圖、街道圖、衛星航空照或立體地形 圖等以辅助使用者獲取適當之多媒體資訊。Information Agent (PIA) and other eight, its work and operation process is as follows: (1) User interface agent: to distinguish which type of device is used by the client, when the user is connected to the line, 'will give appropriate operation Interface and content; _ (2) Monitoring agent: monitor the age of each side-side client, if someone has a new link to inform the user interface agent, give customized information (Cust〇mized); (3) Rain Information Agent: Monitoring The information of the rain station is available to all users at any time; (4) News information agency: news information on the Internet (such as γΑ_奇摩-新闻) automatically grabs the earth and stone flow related (4), the end of the system to see the recent emergency (5) Emergency notification agent: If a certain area is inferred to have a landslide event, it will be returned to all or specific users located in the county area, and the relevant data and situation will be notified; (6) Pre-download agent · User can be located Area to download rainfall or news related information 'If there is an update, the agent will automatically transfer the new data. (7) The agent and the bribe are used to collect the Wei, avoiding the _ information in (5) the server and the client repeatedly access; () SU Belle agent · assist in the layer information download, and give different layers according to their equipment The construction of 4', set of far-including contour maps, street maps, satellite aerial photographs or stereoscopic topographic maps to assist users in obtaining appropriate multimedia information.

11 1284865 於多媒體應用伺服器中,八個智慧型代理其合作之流程圖如第六圖所 示,並將其詳細步驟述說如下 ⑴當使用者210透過網路通訊系統211連接到智慧型代理⑴里,再透過 智慧型代理21裡咖量資訊代理214與糊資訊代理215從雨量資訊 站與新咖211抓取摘有網頁之原始檔,透過事先定義之資料字典 (Data Dictionary)製作出自製之篩選器(Filter)將與本文内容無關之 HTML、ASP或JavaScript等相關標籤(Tags)剔除,加入時間標籤⑺肥 Stamp)、地區、標題(Titles)、全球資源定位||(ijnifQnn Res〇urce11 1284865 In the multimedia application server, the flowchart of the cooperation of the eight intelligent agents is shown in the sixth figure, and the detailed steps are as follows: (1) when the user 210 connects to the smart agent through the network communication system 211 (1) In the past, through the smart agent 21, the amount of information agent 214 and the paste information agent 215 grab the original file of the web page from the rain information station and the new coffee 211, and make a self-made through a predefined data dictionary (Data Dictionary). The filter removes related tags such as HTML, ASP, or JavaScript that are not related to the content of this article, and adds time stamps (7) fat stamps, regions, titles (Titles), global resource positioning||(ijnifQnn Res〇urce

Locator, URL)並與内文(Content)—同交由協調代理218判斷是否為重 覆資訊。 (2) 當協調代理218得到雨量資訊代理214與新聞資訊代理215所蒐集來 之資訊,會依所得之時間標籤、地區、標題與URL判斷是否為重覆資訊。 若非重覆資訊則將此訊息交由緊急通報代理216判別是否有土石流或豪 雨等關鍵字眼。 (3) 若緊急通報代理216發現在新聞文章中已有土石流或豪雨等關鍵字則 在儲存之同時針對該地區發出警訊通知,若無則單純儲存至資料庫中。 (4) 當使用者210登入系統中,會經過監控代理213進行使用者資訊之註 冊(Registration),並獲取使用者之GPS定位值與使用者設備資訊,並 轉使用者e又備> 訊予使用者介面代理212,給予客製化(Customized)的 使用資訊。 (5) 最後使用者介面代理212將相關之參數轉送給予圖層資訊代理219與 12 1284865 ’ 預先下載代理217進行資訊之下載。 (一)路徑預測法(Location-aware Path Predict Method, LP2M) 本系統所採用之地理資訊系統為網格式(Raster)地型圖以向量 式(Vector)資料整合(套疊)方式進行圖層之呈現。在地形圖處理方面先 利用網格式切割法,各類別地形圖層依一定比例之經緯度值將其切分成各 個具一定規則之方格,而當中每一個方塊格在地理資訊資料庫中,皆有建 馨立其空間資料(Spatial Data)與屬性資料(Attribute Data)之關聯性,其 空間資料與屬性資料如下第七圖與第八圖所示,透過系統自動轉換與圖形 >料比對使用者便可以在空間圖形區域中查詢到該區塊所擁有之屬性資 料,亦可由某一屬性資料條件進行查詢符合相同或相似屬性條件之空間圖 形區域為何。經過網格式地形圖形切割法後,當使用者移動所在位置時或 呼叫某區域時,只需下載該區域或週圍區域之地形圖即可。此外使用者在 移動時,無需重新載入一整張地形圖,只需載入變更部份之網格,如此之 •没計可有效地降低網路資料傳輸量,並可降低資料展示之延遲時間(Delay time) 〇 由於使用者利用手持式設備因展示螢幕較小,故在系統中設計能降低 傳輸資流量之圖層資訊代理(Plate Information Agent,piA)下載少量之 週圍區地形圖以利於圖形之展示與降低網路之傳輸。系統中若單純建立每 個區塊所對映之屬性資無法精準標利各個特殊定闕座财訊及地 理資訊,因此RMD2PA在系統所設計之GIS中,亦有存放某部份座標之屬性 13 1284865 資料,系統可透過其座標之屬性資料,劃出向量式之相關圖示。另外本系 統採用將向量式資料和網格式資料相互疊合方式,進行展現地理資訊,其 組成結構是由網格式地形圖建構而成的,而其上之躲標示點與圖像是利 用向量式資料疊套而成的。 在地形圖的呼叫方面,圖層資訊代理(piate Inf〇rmati〇n八咖,piA) path prediCt Method, lp2m) 以順利進行圖層預先下載之工作。由於每個地形方格其規格幾本上是固定 籲的,因此當使用者欲查詢某特定區域之地形圖時,系統之圖層資訊代理 (Plate Information Agent,PIA)會自動依照使用者欲展示之圖層類別、 座標位置等相關資訊,透過圖形内插法與圖形外插法相互配合,即可精確 地呼叫該區域範_之地糊以及週圍區域地糊,並將自動將所有地形 圖區塊組合成一完整地形圖,其運作方式說明如下: 本系統利用網格式切割法進行地形切割,其每區塊方格皆為固定大 小,因此使用者欲查询呼叫某一區域,系統只需取得使用者所欲查詢區域 _之座標,即可推算出該位置座落於哪個方塊,並分析出應下載哪些地形圖 影像。以RMD2PA本系統實作為例,本系統每一區域為長寬分別各占經緯度 5分,此外各個區塊之地圖形檔名依照該區塊之經緯度命名,%「經12〇度 36緯22度36分· gif」或「經12〇度41緯22度%分.gif」等再利用公 式(int((經度-1)/間距)*間距)+ i,即可取得使用者位於哪一個區 塊,將該區域視為中心點,再推算出周圍應置放哪些地形圖。其外插法推 算方式為第九圖所示,圖中之中心點為使用者定位位置,祕塊即為中心 1284865 區塊’接著推算出週圍8個方格,取出後將其整合為-張完整地形圖。不 過由於經緯度為六十進位,每六十分為一度,因此piA需再做檢查動作, 以確保資料正確無誤,PIA以此方式進行地形圖呼个無需再透由資料庫中 的資料來進行搜尋比對,如此可讓雜祕之反應_較雄之切圖法更 有效率。 在-般與網際網路結合之地理資訊系統,其所面臨之重點問題即是網 路傳送資料的鱗,大部份地理資訊祕,#制者移麟—定範圍之外 或剛開始㈣統時都需載人龐大的地糊影賴,雖鮮、統_網格式地 型切割解決單—地形過大之問題,但若在計算能力_手持式設 備及無線通訊網路有限之頻寬下,下載地形圖仍相當耗時,因此難免會出 現延遲的情形。因此本系統設計利用由速度(Vel〇city,〇與加速度 (Acceleration, a)所改良設計之路徑預測法(L〇cati〇n—awareThe Locator, URL) and the content-same are determined by the coordinating agent 218 to determine whether it is a repeating message. (2) When the coordination agent 218 obtains the information collected by the rain information agent 214 and the news information agent 215, it determines whether the information is repeated based on the obtained time stamp, region, title, and URL. If the information is not repeated, the message is sent to the emergency notification agent 216 to determine whether there is a keyword such as a landslide or a rainy day. (3) If the emergency notification agent 216 finds that there are keywords such as landslides or heavy rain in the news article, it will send a warning notice to the area while storing, and if not, simply store it in the database. (4) When the user 210 logs into the system, the monitoring agent 213 performs registration of the user information, and obtains the GPS positioning value and the user device information of the user, and transfers the user e and prepares the message. The user interface agent 212 is given customized usage information. (5) The final user interface agent 212 forwards the relevant parameters to the layer information agent 219 and 12 1284865 ' pre-download agent 217 for downloading the information. (1) Location-aware Path Predict Method (LP2M) The geographic information system used in this system is a Raster map. The vector is presented in a vector (Vector) data integration (stacking) manner. . In the topographic map processing, the net format cutting method is first used. Each type of topographic layer is divided into squares with certain rules according to a certain proportion of latitude and longitude values, and each of the squares is built in the geographic information database. The relationship between the spatial data and the Attribute Data is as follows. The spatial data and attribute data are as shown in the seventh and eighth figures, and the user is automatically converted and graphed by the system. The attribute data owned by the block can be queried in the space graphic area, and the spatial graphic area of the same or similar attribute condition can be queried by an attribute data condition. After the mesh terrain graphics cutting method, when the user moves the location or calls an area, simply download the topographic map of the area or surrounding area. In addition, when the user is moving, there is no need to reload an entire topographic map, and only the grid of the changed part is loaded, so that the amount of network data transmission can be effectively reduced, and the delay of data display can be reduced. Delay time 〇 Because the user uses the handheld device because the display screen is small, the Layer Information Agent (piA) is designed to reduce the transmission traffic. The terrain information map is downloaded to facilitate the graphics. Show and reduce the transmission of the network. In the system, if the attributes of each block are simply not able to accurately target the special financial information and geographic information, the RMD2PA also has the attributes of some coordinates in the GIS designed by the system. 1284865 data, the system can draw a vector-like icon through the attribute data of its coordinates. In addition, the system adopts the method of superimposing the vector data and the network format data to display geographic information, and the composition structure is constructed by the network format topographic map, and the hiding points and images thereon are using vector formula. The data is stacked. In the topographic map call, the layer information agent (piate Inf〇rmati〇n eight coffee, piA) path prediCt Method, lp2m) to smooth the layer pre-download work. Since each terrain square has a fixed number of specifications, when the user wants to query the topographic map of a specific area, the system's Layer Information Agent (PIA) will automatically display the user according to the user. Relevant information such as layer category and coordinate position can be accurately matched by the graphics interpolation method and the graphic extrapolation method, and the local area and the surrounding area can be accurately called, and all the topographic map blocks will be automatically combined. The complete topographic map is described as follows: The system uses the mesh format cutting method for terrain cutting, and each block has a fixed size. Therefore, the user only needs to obtain a user to query a certain area. To query the coordinates of the area _, you can calculate which square the location is located in, and analyze which topographic map images should be downloaded. Taking the RMD2PA system as an example, each area of the system has a length and width of 5 points each according to its longitude and latitude. In addition, the graphic file name of each block is named according to the latitude and longitude of the block, and the percentage "is 12 degrees and 36 degrees latitude 22 degrees. 36 points · gif" or "after 12 degrees 41 latitude 22 degrees % points. gif" and other reuse formula (int ((longitude -1) / spacing) * spacing) + i, you can get the user located in which area Block, consider the area as the center point, and then calculate which topographic maps should be placed around. The extrapolation method is shown in the ninth figure. The center point in the figure is the user's positioning position, and the secret block is the center 1284865 block'. Then the surrounding 8 squares are calculated, and then taken out and integrated into - Zhang Complete topographic map. However, since the latitude and longitude is 60 decimal, every six tenths is one degree, so the piA needs to check again to ensure that the data is correct. The PIA does not need to search through the data in the database. Comparison, this can make the reaction of the mystery _ more efficient than the cut method. In the geographic information system that is integrated with the Internet, the key issue is the scale of the network transmission data, most of the geographic information secrets, #制者移麟- outside the scope or just beginning (four) At the same time, it is necessary to carry a huge amount of ground-breaking shadows. Although the fresh, unified _ mesh format cutting solves the problem of single-over-large terrain, but if the computing power _ handheld devices and wireless communication network limited bandwidth, download Topographic maps are still quite time consuming, so delays are inevitable. Therefore, the system design utilizes the path prediction method (L〇cati〇n-aware) designed by velocity (Vel〇city, cel and acceleration (Acceleration, a)).

Predict Me1:hod,LP2M)以進行使用者之行進路徑預測,當系^預測到使用 者之行徑之下一個區域為何處時,即會預先下載該區域之地形圖至使用者 没備之記憶裝置内,如此當成員實際進入該區域時,即可立即調出該區域 之地形圖,如此即可避免傳輸展示延遲(Delay)之情形,其路徑預測法之預 測公式如下:Predict Me1:hod, LP2M) is used to predict the travel path of the user. When the system predicts the area under the user's path, the topographic map of the area is pre-downloaded to the memory device that the user has not prepared. Therefore, when the member actually enters the area, the topographic map of the area can be immediately called up, so that the transmission delay (Delay) can be avoided. The prediction formula of the path prediction method is as follows:

At = t1 - t2 …式(1) v =( p(t) - p(f ))/At …式(2) a = Λν / At …式(3) 15 1284865 使用者路徑刪主要分析要素包括使用者目前位置、前一時間位 置_,)與間格隔時間^等三個要素。當系統取得使用者目前位置_ 與前-個時間點位置〆〇時,利用平均速度功式求出使用者目前行徑平 均迷度f為多少,並再利用兩個時間點之平均速度之差,也就是兩個時間 點之平速度變化量^,來取得平均加速度a為何。由於加速度具有大小和 方向的物理量’其使用者行徑之變換與加速度變化量有密切關係,由此可 知我們可利用此性質進行使用者行徑路徑鋼。因此本發明將依平均加速 •度3與行徑模式之間的關系,可分為下列二項: ⑴當平均速度V為V’時,表示成員路行徑為正向行進,並又可分為: (i)平均加速度a = 0 此表示使用者在㈣歧段時間,平均速度並沒改變,即㈣,此時^, 這段時間内做等速度運動,因此我們可預測使用者短時間内會朝著同一方 向前進,因此可預先下载下個目的地之地形圖影像。 (ii)平均加速度a > 0 .此表示使用者在㈣這段時間,有加速之情形,使用者應會朝同一方尚 前進’但由於平均速度加快,表示使用者將更快速達到下—個目的區塊, 因此需更提早預先下載下個目的地之地形圖。 (iii)平均加速度a < 0 表示使用者速度到已慢慢減緩的趨勢,這表示使用者有可能即將停止,或 進行轉向’因此_絲灯鄕像,或下似财向之地形圖。 ⑵當平均速度v為時,表示其路行徑為反向行進,而利用平均加速 .1284865 度也可分為三種情形,不過條件所得的結果會^為‘v,,時相反,平均 加速度a >卜則表示即將停止或進行轉向,而若平均加速度a < 〇則 表式有反方向加速之情形。 如第十圖所示,目前使肖者所處中心位置為MapP之區塊,在使用者由价,) 移動至ρ(ί,)時LP2M預測使用者下一個最有可能的中心位置Map卜因此_ 即預先下載Map2、Map3、Map4賴塊之地糊至伽者設備巾。當使用者移 至Map卜即可立即調出該區塊之地形圖,讓使用者減少等待地形圖下載之時 間。在行進方向與下_層之判定可由所行進之定位點L ϋ,⑴與&進行 判斷,若為上下左右絲示使用者行進方向為直線,而為圖之四肖表示其行逕 方向為斜向,由此可判斷所需下細片之數目。第11圖為直向行走則需預先 下載3張圖,而第12圖為斜向行走則須預先下載5張圖,第丨丨圖與第12圖 上之數字卜2、3、4,5表示下載之順序,虛線框表示畫面所展示之圖像。 (三)虛擬實境設計 虛擬實境(VR)之設計是讓在現地探察的使用者,或位於遠處的專家能 在任何地點使用此魏查看各區域之相關地形圖與虛擬實境災情模擬展 示。由於祖程式語言產生之程式碼標案較小,故有利於在GpRs上傳送 相關之檔案與模擬情況。針對較大或較精確的地形,運算綠圖仍極耗費 系統資源’因此利用平板電腦開啟較大的舰形圖時,其所花費處理時間 也相對地較長,且不能像在PC端—樣的快速赫視點,有時甚至出現延遲 展示之情形。另料板電腦螢幕展轉減麟視畫面較桌上型電腦差, 17 .1284865 基於上述理由本系統利用ERDAS IMAGINE8.5軟體,將全台灣的立體地形圖 依經緯度分別切割成多個小區塊,並把較常發生土石流與發生機率較高之 地區獨立切割出來。在分割同時也藉由該軟體將各區塊之地形解析度及格 數調高,唯各區塊之經緯度必須紀錄在後端的資料庫中,因此前端之平板 電腦使用者利用GPS所取得當地經緯度或輸入欲查看區域座標時,系統即 會尋找該區域座落在哪個區塊中,並直接呼叫出該區塊的VR地形檔,因此 圖形較小且切割出之圖形精緻度提高,此方式即可提供解析度高並且較不 失真的立體地形給使用者參考,故在PDA上能展示出較佳之效果與效能。 (四)Web化服務設計At = t1 - t2 (1) v = ( p(t) - p(f )) / At ... (2) a = Λν / At ... (3) 15 1284865 User path deletion main analysis elements include The user's current location, the previous time position _,) and the interval time ^ and other three elements. When the system obtains the user's current position _ and the previous time point position ,, the average speed power is used to find out the average average f of the user's current path, and the difference between the average speeds of the two time points is used. That is, the amount of change in the flat speed at two time points ^ to obtain the average acceleration a. Since the acceleration has a physical quantity of magnitude and direction, the change of the user's path is closely related to the amount of acceleration change, and thus it can be known that we can use this property to perform the user path path steel. Therefore, the present invention will be divided into the following two items according to the relationship between the average acceleration degree 3 and the path mode: (1) When the average speed V is V', it indicates that the member path is positively traveling, and can be further divided into: (i) Average acceleration a = 0 This means that the average speed has not changed during the (4) segment time, that is, (4), at this time ^, during this time, the equal speed motion is performed, so we can predict that the user will be in a short time. Moving in the same direction, the topographic map image of the next destination can be downloaded in advance. (ii) Average acceleration a > 0. This means that during the period of (4), there is acceleration in the user, and the user should advance toward the same side. 'But the average speed is faster, indicating that the user will reach the next faster. For the destination block, it is necessary to pre-download the topographic map of the next destination in advance. (iii) The average acceleration a < 0 indicates the tendency of the user's speed to slowly slow down, which means that the user may be about to stop, or to turn to the 'thus' image of the silk lamp, or the topographic map of the financial direction. (2) When the average speed v is, it means that the path is reversed, and the average acceleration of 1.2844865 degrees can be divided into three cases, but the result of the condition will be ^v, and vice versa, the average acceleration a &gt Bu indicates that the steering will be stopped or turned, and if the average acceleration a < 〇 then the situation has the opposite direction of acceleration. As shown in the tenth figure, at present, the center position of the viewer is MapP, and when the user moves from price to ρ(ί,), LP2M predicts the user's next most likely center position. Therefore, _ pre-downloads the map, Map3, and Map4 blocks to the gamma equipment towel. When the user moves to the Map Bu, the topographic map of the block can be called up immediately, allowing the user to reduce the waiting time for the topographic map to be downloaded. The determination of the traveling direction and the lower layer can be judged by the traveling positioning points L ϋ, (1) and & if the upper and lower left and right wires indicate that the user's traveling direction is a straight line, and the four-dimensional diagram of the figure indicates that the traveling direction is oblique Thus, the number of required lower patches can be determined. Figure 11 shows that three pictures need to be downloaded in advance for straight walking, and five pictures are required to be downloaded in advance for the diagonal walking. Figures 2 and 3, 4, 5 on the second and fourth pictures. Indicates the order in which the downloads are made, and the dashed box indicates the image displayed on the screen. (III) Virtual Reality Design Virtual Reality (VR) is designed to allow users who are in the field to explore, or experts located in the distance to use this Wei to view the relevant topographic maps and virtual reality disaster simulations of each area. Show. Since the code generated by the ancestor language is small, it is advantageous to transmit related files and simulations on GpRs. For larger or more accurate terrain, computing green maps still consumes a lot of system resources. So when using a tablet to open a large chart, the processing time is relatively long, and it can't be like on the PC side. The fast point of view, sometimes even the case of delayed display. In addition, the computer screen display of the board is worse than the desktop computer. 17.1284865 Based on the above reasons, the system uses the ERDAS IMAGINE8.5 software to cut the three-dimensional topographic map of Taiwan into several blocks according to the latitude and longitude. Cut out the more common occurrences of earth and rock flow and the areas with high probability of occurrence. At the same time, the software also increases the topographic resolution and the number of cells of each block by using the software. Only the latitude and longitude of each block must be recorded in the back-end database, so the front-end tablet user obtains the local latitude and longitude by using GPS. When you want to view the area coordinates, the system will find out which area the area is located in, and directly call out the VR terrain file of the block, so the graphics are smaller and the cut graphics are improved. Provides high resolution and less distortion of the stereoscopic terrain for the user's reference, so it can show better results and performance on the PDA. (4) Web service design

Web化服務是提供使用者利用終端設備連上多媒體應用飼服器,所提供 之晝面依設備處理之不同’智慧型代理會自動調適其糊晝面傳送資料至 使用者端,此外亦提供大量土石流相騎聞與資訊供使用者進行查詢與參The Web-based service provides users with a terminal device to connect to the multimedia application server. The provided information is handled by the device. The smart agent automatically adjusts its paste to transmit data to the user. In addition, it provides a large number of services. Earth and stone flow riding and information for users to query and participate

考。使用者若漫遊(Roaming)到無GPRS基地台服務之地區時,則可使用娜 系統之WAP服務。在多媒體應用伺服器中亦提供Mp/WEB的應用,經由互 動式網頁(ASP)技術與無線標註語言(狐)存取會議内容。因譬手機的功 能與頻寬限制,僅能接收與解讀wb即的圖像格式,故在此系統中翻 手機的微觀H (Mi⑽—bnDwser)進行衫的即時傳輸與接收。test. If the user roams to an area where there is no GPRS base station service, the WAP service of the system can be used. The Mp/WEB application is also provided in the multimedia application server, and the conference content is accessed via the interactive web page (ASP) technology and the wireless annotation language (fox). Due to the function and bandwidth limitation of the mobile phone, only the image format of wb can be received and interpreted. Therefore, in this system, the micro H (Mi(10)-bnDwser) of the mobile phone is used for instant transmission and reception of the shirt.

Web伺服器亦提供提供案例式細丨擎,讓制者可由終端設備輸入相 關資訊進行土石流案例之推估。因此細者並不需要—定為本系統使用 者,其他使用者均可一起使用。 1284865 三、專家決策系統 位於網際網路後端之專家決策系統,主要工作為架構相關之推理引 擎,在此設計案例式推理引擎進行建置土石流預測模式,經由驗證與比較 後得到之模式建置於行動化土石流即時預測系統中,提供行動化使用者進 行災前即時決策判斷之參考。 _ (一)土石流危險因子之選定 針對民國89年南投縣境内具15度以上有效集水面積,所劃定之ΐ8ι 條土石流潛勢溪流,並以民國90年桃芝颱風後,發生於此181條潛勢溪流 上之土石流災害數據資料做為研究樣本。土石流危險因子之選定為土石节 即時預測祕中最重要的-環,正確_子選定可使糊模式之規劃與分 析達到纖的效果,進而準確酬土石流災害之發生。造成土石流發生之 三大重要條件:⑴足夠的水供應、⑵足夠的溪床堆積物與材料來源、 ®⑶足觸坡度條件。 而本發明所選定之危險因子分為非即時性因子與即時性因子:(1)非 即時性因子:為災害環境中經由長期環境變化得來之危險因子,需以長期 性監測找出此因子變化情形。非即時性因子包括⑴有效集水面積、⑴) 有效溪床長度、㈤)有效溪床坡度、(iv)有效集水區内岩性。⑵即時 性因子:為災害魏中即時變化所產生危關子,包括⑴有 崩塌面積、(u)有效累積雨量、(出)有效降雨強度與(iv)植生指數等 .1284865 即時性因子。其中除了纽_面觀子充滿不確定性外,有效累積 與有效降爾,咖,㈣嫩_概咖代理取得, 並利用即雜因子之訊息’進—步即雜職害魏之變化潛能。 針對有效地區_之計算,本發明為衛星__(Remote S嶋ng,RS)技術,以近紅外光波段⑽與植生指數(職),經地球表面 反射後所制_射餘代有效鮮區_塌面積,因此有賴水區内平 均IR或NDVI值偏低者,表示有效集水區内平均崩塌面積較大。如第^圖 所示’以20料南投災區SP〇T魅影像轉換與分類統計後可得到植生= 與崩塌裸露地之分界線。 (一)土石流危險因子之粹取 土石流危險因子粹取之精確度為回歸模式成敗的重要因素,粹取不良 的因子變化值將會成為回歸模型眾多數據中的“雜訊,,,甚至造成回歸分 析結果與事實不符,導致回歸模型可信度降低。因此本發明運用㈣奶資 磁術進付際駐石流危險因子粹取,以建置南娜土石流危險因子資 料庫。本發明所選定之八個土石流危險因子粹取方法分述如下。 (1)非即義因子之獅··本發明所選定的非即時性因子包含有效集水面 積、有效溪床長度、有效溪床坡度與有效集水區内岩性等四個因子,由 於這些危險因子屬於天然災害環境中較不易改變的環境因子,須經由長 期性觀測且需較複雜之地形計算,方能進行土石流危險因子之粹取。因 此本發明應用GIS與RS資訊技術,設定15度以上集水面積為該流域之 20 .1284865 .-·有效集水面積,第15圖所示為所粹取出之有效集水面積範圍,第關 鮮17 麵套蝴,物麵編_水面積、 溪床長度、溪顧度與岩_解危_子資訊,以作為數值回歸分析 之依據。 (ω㈣性因子:本發衝選定的㈣性奸包含有效累積雨量、有效降 雨強度、有效集水區《紅外光⑽平均值財效集搞喊生指數 (_)平均值等四_子,其中有效累積雨量(麵)與有效降雨強度 # Uhr/咖〉’可以從_F)2系統中之環境認知代理即時取得雨量即時資 訊。 (三)土石流預測模式理論與架構設計 本發明所提出之土石流災害麵模式為利用多元線性回歸、多變量不安 定指數法細_貞神經網路等分析方法所建置,侧研究理論與研究架 構設計分述如下。 (1)多元線性回歸分析法 在多元線性回歸方法中,我們設定變數k期望值,為自變數阳十 2、…、幻之線性函數,且依獨立賴誤差變數匕調整誤差精確度,因此 多元線性模型假設為:The web server also provides a case-by-case engine that allows the system to input relevant information from the terminal device for the estimation of the earth-rock flow case. Therefore, the latter is not required—it is intended for use by the system, and other users can use it together. 1284865 3. The expert decision-making system is located in the expert decision-making system at the back end of the Internet. The main work is the architecture-related reasoning engine. Here, the case-based reasoning engine is designed to build the prediction model of the earth-rock flow, and the model is obtained after verification and comparison. In the mobile land flow real-time forecasting system, it provides a reference for mobile users to make immediate decision-making before the disaster. _ (1) The selection of the risk factor for earth-rock flow is based on the effective catchment area of 15 degrees or more in Nantou County in the Republic of China in 1989. The 8 条 土 土 土 土 , , , , , , , , , , , , 181 181 181 181 181 181 181 181 181 181 181 181 181 Data on earth-rock disasters on the potential streams are used as research samples. The selection of the risk factor of earth-rock flow is the most important---the ring of the earth-rock festival. The correct-sub-selection can make the planning and analysis of the paste mode reach the effect of fiber, and then accurately compensate for the occurrence of earth-rock disaster. Three important conditions for the occurrence of earth-rock flow: (1) adequate water supply, (2) sufficient bedbed deposits and material sources, and (3) foot-slope conditions. The risk factors selected by the present invention are classified into non-immediate factors and immediate factors: (1) Non-immediacy factors: risk factors derived from long-term environmental changes in a disaster environment, which need to be identified by long-term monitoring. Change situation. Non-immediacy factors include (1) effective catchment area, (1) effective pond length, (5) effective river bed slope, and (iv) lithology in the effective catchment area. (2) Immediateness factor: It is a crisis for the immediate change of disasters in Weizhong, including (1) collapse area, (u) effective accumulated rainfall, (out) effective rainfall intensity and (iv) planting index, etc. 1284865 Immediateity factor. In addition to the New_face view is full of uncertainty, effective accumulation and effective demotion, coffee, (four) tender _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ For the calculation of the effective area _, the present invention is a satellite __ (Remote S嶋ng, RS) technology, in the near-infrared light band (10) and the vegetative index (career), after the reflection of the earth's surface, the _ ray generation effective fresh area _ The collapse area, therefore, depends on the average IR or NDVI value in the water zone, indicating that the average collapse area in the effective catchment area is large. As shown in the figure below, the boundary between the planting and the collapsed bare ground can be obtained after the conversion and classification of the SP〇T image of the 20-nannan disaster area. (1) The risk factor of the earth-rock flow risk is taken as the important factor of the success or failure of the regression model. The value of the negative factor change will become the "noise," and even the regression in the regression data. The results of the analysis are inconsistent with the facts, resulting in a decrease in the credibility of the regression model. Therefore, the present invention uses (4) the risk factor of the stone flow in the milk magnetic charge to establish a database of risk factors for the Nana soil flow. The eight methods for extracting dangerous factors of earth-rock flow are described as follows: (1) Lions of non-immediate factors · The non-immediate factors selected by the present invention include effective catchment area, effective bed length, effective riverbed slope and effective set Four factors, such as lithology in the water zone, are important environmental factors that are difficult to change in natural disaster environments. They must be calculated through long-term observations and complex terrain calculations in order to extract the risk factors of soil and rock flow. The invention applies GIS and RS information technology, and sets the catchment area of 15 degrees or more to the water level of 20.1284865.-·effective water collecting area, Figure 15 For the range of effective catchment area that was taken out, the first 17 sets of the first set of butterflies, the surface area _ water area, the length of the river bed, the Xi Gudu and the rock _ _ _ sub-information, as the basis for numerical regression analysis. (ω (four) factor: The selected four (4) sexual traits contain effective accumulated rainfall, effective rainfall intensity, effective catchment area, "infrared light (10) average financial effect set, shouting index (_) average, etc., among them Effective cumulative rainfall (face) and effective rainfall intensity # Uhr/咖〉' Instantly obtain rainfall real-time information from environmental awareness agents in the _F)2 system. (III) Earth-rock flow prediction model theory and architecture design The earth-rock flow proposed by the present invention The disaster surface model is constructed by using multiple linear regression, multivariate instability index method, 贞 neural network and other analytical methods. The side research theory and research architecture design are described as follows: (1) Multiple linear regression analysis in multiple linearity In the regression method, we set the expected value of the variable k, which is the linear function of the self-variable yang, 2, ..., magic, and adjust the error precision according to the independent error variable ,, so the multivariate linear model assumes:

Yl = B。+ BlXll + B2X12 + ··· + BkXlk + ε J 21 (S ) 1284865 Ϊ2 = Bo + Β1Χ21 + Β2Χ22 i ··· + BkX2k +£2 φ Φ 0Yl = B. + BlXll + B2X12 + ··· + BkXlk + ε J 21 (S ) 1284865 Ϊ2 = Bo + Β1Χ21 + Β2Χ22 i ··· + BkX2k +£2 φ Φ 0

Yn = Bo ~l· BlXnl + B2Xn2 + - + BkXnk +£n 其矩陣表示式如下: 1 工11 Χ12 Λ - Λ' V 一 1 X21 ^22 八 Xu A ^2 Μ Μ Μ Μ μ Μ + Μ 人 -1 Ή”2 Λ Λ. 正規方程組之轉換則以最小二乘法及偏微分後得到:Yn = Bo ~l· BlXnl + B2Xn2 + - + BkXnk +£n The matrix representation is as follows: 1 work 11 Χ12 Λ - Λ' V a 1 X21 ^22 八Xu A ^2 Μ Μ Μ Μ μ Μ + Μ person -1 Ή"2 Λ Λ. The conversion of the normal equations is obtained by least squares and partial differentiation:

nb〇+biExi+b2Ix2+ ... +bkIxk=IY boZxi+biExi2 +b2Exi.X2+ ... ihZxi^Xk=Zxi^YNb〇+biExi+b2Ix2+ ... +bkIxk=IY boZxi+biExi2 +b2Exi.X2+ ... ihZxi^Xk=Zxi^Y

b〇Exk+biExi^xk+b2Ex2^xki ··· +bkExk2=Exk^Y 因此由上式可知,以上正規化方程組中之係數矩陣為對稱矩陣,令j 代表係數矩陣,而B則為右端常數項矩陣,矩陣方程式如下所八 Σ〜λ Έχι T,xrx2^Ydx1-xk Μ Μ μ -1 1 h ^21 Λ Μ Μ Xjk 〜Λ xnk Ί χπ Χ12 Xlk 1 Χ21 Χ22 X2k Μ Μ Μ 1 Λ ' xf · χ s 22 1284865 β 1Γ Σ^b〇Exk+biExi^xk+b2Ex2^xki ··· +bkExk2=Exk^Y Therefore, we can see from the above equation that the coefficient matrix in the above normalized equations is a symmetric matrix, so that j represents the coefficient matrix, and B is the right end. The constant term matrix, the matrix equation is as follows: gossip ~ λ Έχι T, xrx2^Ydx1-xk Μ Μ μ -1 1 h ^21 Λ Μ Μ Xjk ~Λ xnk Ί χπ Χ12 Xlk 1 Χ21 Χ22 X2k Μ Μ Μ 1 Λ ' Xf · χ s 22 1284865 β 1Γ Σ^

Γ /XΓ /X

AA

XX

X 7 ^ Λ 2 λλ^ 2 ΑΛ Λ 因此正規化方程式的矩陣型式為公式(4)X 7 ^ Λ 2 λλ^ 2 ΑΛ Λ Therefore, the matrix type of the normalized equation is the formula (4)

(^/*^)b=x/T Ab=B .式⑷ ”b=[b°、bl、…、w為正規方程式的未知數。由於在係數矩陣滿秩 的條件下存在Oc’.x;)的逆矩陣,所以係數&可由以公式(4)可求得: b-Cx^x) xf .γ .式⑸ 再經由公式(5)可推得多元線性模型為下式(6) ·· Y = b〇 + blXl + b2X2 + bsX3 i - + bkXk …式(6) (2)多變量不安定指數分析法 多變量不安繼編,糊輪各咖_子之變異性, 與估算土石崎物錢版鳩峨其影響 23 1284865 比例广予各因子不同權重之評分值,進而推導可適性高之統計計量評估 數學模式,及計算咖子分類之不安定指數其中λ條卜㈣之相 對等級評分值,細子之累積值越接㈣表示土石流發生的潛在機率越 南,反之,則發生潛在機率越低。从定義如公式(?):(^/*^)b=x/T Ab=B . (4) ”b=[b°, bl, ..., w are unknowns of the normal equation. Oc′.x exists due to the full rank of the coefficient matrix; The inverse matrix, so the coefficient & can be obtained by the formula (4): b-Cx^x) xf .γ . (5) Then the formula (5) can be derived to obtain the multivariate linear model as the following formula (6) · Y = b〇+ blXl + b2X2 + bsX3 i - + bkXk ... (6) (2) Multivariate instability index analysis multivariate restless succession, paste wheel coffee _ variability, and estimation of earth ashi The money version has its influence 23 1284865. The ratio is widely distributed to the scores of different weights of each factor, and then the mathematical model of statistical appraisal with high applicability is derived, and the instability index of the classification of the coffee category is calculated, among which the relative rank score of λ (4) The value, the cumulative value of the thinner (4) indicates the potential probability of occurrence of the earth-rock flow in Vietnam, and vice versa, the lower the potential probability occurs. From the definition as the formula (?):

Drd>d>d>^d:n ···式(7) Λ為潛在危險度之因子分類不安定指數 春di,…,ά為各變異影響因子之評分值 奶,…,脱為各變異影響因子之權重值 各變異因子之分級級數評分值計算方法定義如下公式(8): a{Xt - X \ =(^^+7 …式⑻ a為特定常數值 A為各變異影響因子之評分值 尤為土石流之各因子分級發生百分比 I與I分別為尤之最大及最小值 各單—因子之變異係數計算方法定義如下式(9): 24 1284865 χ xI〇0% …式(9) 與σ分別為變異係數與標準偏差 為各因子分級之破壞百分比平均值 子之變異係數除以全部因子之變異係數總和,即為該^^ 直敫權重值,定義如式(10): …式(10) υ/+υ2+Λ +υ„ 的為該因子之有效權重值 i為各因子之變異係數 (3)倒傳遞類神經網路分析法 > 類神經網路是一種基於腦與神經系統研究所啟發的資訊處理技術。它 可以利用系統輸出與輸入資料來建立系統模型,此種類神經網路系統模型 可用於推估、預測、決策、診斷等應用範圍,因此從廣義的角度來看,類 神經網路也是一種特殊型式的『非線性統計技術』。本發明採用最具代表性 及應用最為普遍的倒傳遞類神經網路演算法分析土石流潛在危險度。該演 真法為一典型監督式學習網路(Supervised Learning Network),主要學習 輸入與輸出間之内在對映規則,而此規則以各網路處理單元連接加權值表 25 1284865 示。因此未來有新的案例產m由輸人值或自變數值即可得到推論 相關輸出值,因此本發_關傳遞類神經演算法預測土石流潛勢溪流^ 危險程度,與土石流發生之風險概率。 (4)資料前處理分析 各影響因子資料於倒傳咖神_路紐進行姆前,賴對資料作 前處理,使得财賴的輸人魏映射至__。因監督摘傳遞類神 卜經網路之輸人處理單猶射接受㈣大小值,但如果獨處理單元 所接受的魏值錢社大,將使小鶴魏之重概無絲現出來,造 成大值域變數控制整個網路學習過程,而影響網路學習效果。因此本發明 所採用之輸入變數值域轉換方式為機率對映法,所設計之研究步驟論述如 下。⑴統計變數的平均值A與鮮差α,⑵設定所需的最小值為L 與最大值/)_,(3)資料規袼化公式(η)定義如下:Drd>d>d>^d:n ··· (7) Λ is a factor of potential risk classification instability index spring di,...,ά is the score value of each variation factor, milk,..., The weighting factor of the influence factor The calculation method of the graded grade score value of each variation factor is defined as the following formula (8): a{Xt - X \ =(^^+7 (8) a is a specific constant value A is the influence factor of each variation The scores are especially the percentages of the various factors of the earth-rock flow. I and I are respectively the maximum and minimum values. The calculation method of the coefficient of variation of each factor-factor is defined as follows (9): 24 1284865 χ xI〇0% ... (9) σ is the coefficient of variation and the standard deviation are the coefficient of variation of the average percentage of damage of each factor divided by the sum of the coefficients of variation of all factors, that is, the weight value of the ^^ straight, defined as equation (10): 10) υ/+υ2+Λ +υ„ is the effective weight value of the factor i is the coefficient of variation of each factor. (3) Inverted-transfer-like neural network analysis> The neural network is based on the brain and nervous system. Research-inspired information processing technology that takes advantage of system output and input It is expected to establish a system model. This kind of neural network system model can be used for estimation, prediction, decision making, diagnosis and other applications. Therefore, from a broad perspective, neural network is also a special type of "non-linear statistical technology". The present invention uses the most representative and most commonly used inverse transfer neural network algorithm to analyze the potential danger of soil flow. The real method is a typical supervised learning network (Supervised Learning Network), which mainly learns between input and output. The internal mapping rule, which is shown by each network processing unit connection weighting table 25 1284865. Therefore, in the future, there is a new case where the output value can be inferred from the input value or the self-variable value, so the present invention _Off-transfer-like neural algorithm predicts the potential of the landslide flow ^ danger degree and the risk probability of occurrence of the landslide. (4) Data pre-processing analysis of each impact factor data in the back of the gods _ Lu New Zealand, before the data Pre-processing, which makes the input of the financial reliance map to __. Because the supervisor picks up the input type of the gods, the input of the human beings accepts the single-infrared acceptance (four) size value However, if the Wei-Chengshe, which is accepted by the single-processing unit, will make the small crane Weizhi appear out of the way, causing the large-value variable to control the entire network learning process and affect the network learning effect. Therefore, the invention adopts The input variable-value domain conversion method is the probability mapping method. The designed research steps are as follows. (1) The average value of the statistical variables A and the fresh error α, (2) The minimum value required for setting is L and the maximum value /) _, (3) The data specification formula (η) is defined as follows:

(^max — ^min ) ^min …式(11) 其中I為自變數^為平均值;r為標準差;1與1為最小值與最 大值;ic為轉換係數,當技29可將99· _資料映射至[H]區間, 其餘㈣.8卜 2. 58、1. 96、i 65、h 28 分別將 99 5%、99%、9热、_、觀 的變數值映射到[/)— 區間。 本發明所獅之倒傳遞類神經網路資料前處理分析方法,為針對各個 26 1284865 〜響®子’分別求出訓練範機數之平均值與鮮差,以顧機率對映法 公式將每一樣本進行變數值域之轉換。 (5)倒傳遞類神經網路演算流程 倒傳遞網路系統之操作與演算過程可分為以下幾個步驟⑴設定網路 多數’(2)明佈隨機亂數設定加權矩陣i处及【細,與偏權向量❹ 及初始值,⑶計算隱藏層(Hidden㈣的輸出量;⑷設定輸出層 鲁(Output layer}與隱藏層容許差距量:⑸計算輸出層與隱藏層差距量占; ⑹判別輸出層與隱藏層差距量是否大於料差距量,若差距量小於容許 差距1,則可求出最佳回歸模式;⑺若輸出層與隱藏層容許差距量大於 則計异输出層與隱藏層加權矩陣及偏權值修正量;⑻重新修正輸出層與 IU藏層加權矩陣與偏權值,重覆⑶—⑻直到輸出層與隱藏層差距量在容許 差距量中(即不滿足步驟⑹時),並比對敏感度修正量之糊性,以找出最 鲁佳回歸模式。上述之倒傳遞類神經網路演算法流程整理於第Μ圖。 ⑹網路模式之敏感度分析(Sensitivity Analysis) 運用倒傳遞類神經網路時,如果輸人與輸出單元賴係近似單調 (H〇m_ous),則可從連結上的加權分析網路,分別輸入各單元之對映輸 出單元敏感度。敏感度正值越大表示二者之間的正相關性越大,負值越大 表示二者間負相_大。而此過程代表網路學f並以最陡坡降法來使能量 函數最小化,即每當輸入一個訓練範例時,網路則必須小幅調整加權值大 27 .1284865 ' 小’使得調整幅度與誤差函數對該加權值的敏感程度成正比,即誤差函數 - 對加權值的偏微分值大小成正比,上述關係值定義如下式(12): -η 6Ε ~dW~ …式(12) 凡為介於第i、y個處理單元間的連結加權值 ^ ^為學習速率用以控制最陡坡降法之誤差調整幅度 最後將所粹取出之非即時性因子值與即時性因子值傳回後端專家系統 進行即時回歸運算,進而正確即時預測當地土石流潛在危險度。第19圖為 本發明依據南投縣境内181條土石流危險因子粹取值,以關聯式資料庫鏈 結與物件導向語言建立土石流危險因子資料庫。 位於網際網路後端之主動式災情通報資料庫系統,其主要功能為利用 E-C-A模型進行資料主動之查詢與給予,並且負責與WinCE資料庫進行同步 馨之工作此外,針對逃生路徑之選擇亦為重點之一。此發明提出與設計之 功能與方法主要包含主動式資料庫之架構與設計及逃生路徑模擬,其相關 設計方法分述如下。 . (7)主動式資料庫(Active Database) 此發明將設計改良傳統的資料庫成為一個主動式資料庫(Active Database) ’在所設計之主動式資料庫中採用 E-C-A(Event-Condition-Action)模型處理資料,讓所有的資訊不再被動查 28 1284865 f2胃料轉依制者之情況主_知使財執行邊《塌資料查 、、/Γ用# $上料立觸問衫要輸人邊坡崩塌資料 :若有新 的邊坡朋塌«訊料會快速通知制者,Μ讓伽者自己下查詢語法 (㈣17 Lang峰),進而辟邊《戦情通_。此外_者查詢 附近地域之邊坡旦發現使用者正位在危險區時,則此系統 Z刻以警_方式通知伽者,查詢亦由贿触絲财式以達到 最高的成效與承擔最小之災害風險。 Φ邊坡崩狀害⑷兄:貝料庫建置,採用單-技術無法解決全部問題,故 利用-架構進行私相關技術。近年來,由於遙測技術⑽、地理資訊系 統(GIS)、全球衛星定位系統(GPS)與衛星影像處理分析系統(ips)等相關 科技之成熟發展,整合建置邊坡崩塌防治技術與Gps/Gls/Ips科技之串聯 應用,進一步積極發展邊坡崩塌防治技術自動化和資訊化系統,對於後續 水土資源保育利用和邊坡崩塌災害防治,均可提供精確快速之資訊交流互 通,實有益於邊坡崩塌災害環境資料庫作業系統之建構。若就邊坡崩塌災 ^害防治技術應用需求層面而論,該整合系統可分以下三部份: (i) 自然環境:包括行政區域、地形、地質、地貌、水系、流域、土壤、 氣象、水文、植群分佈等相關資料圖層。 (ii) 人文壞境·包括土地利用、開路、採礦、伐木、棄土、濫墾、濫葬、 濫建等人為活動。 (iii) 災害環境:山崩、地滑、土石流、底土滲流、向源侵蝕、土壤沖麵 等潛在性災害。 29 1284865 (8) 逃生路線模擬 當災害發生時,災區道路受損,災民往往呈現一片混亂不知該往何處 逃生’本發明設計一方法稱為最適路徑法(Adaptive Path Method,APD), 進行最適逃生路徑之推論,並透過虛擬實境(virtual Reality,VR)之模擬 展不相關路線,使用者僅須使用pDAM上全球衛星定位系統(Gp幻透過無線 通訊網路(Mobile Network)連結到網際網路上決策支援系統(DSS),以進行 最適逃生路徑之推論。決策支援系統推論模組利用螞蟻演算法(八的 Algorithm)並配合最祕#演算法(ShQrtest祕MethQd)進行推估其 中為有效反應現實災區情況,此發明利用背包問題(Knapsack p⑺_)將 避難所距離、_地可容納人數、存糧鄕療狀況特輯行最適逃生路 徑建議’以自動避免災民過於集中避難地之困境,確實輔助災民逃生。所 利用之螞蟻演算法與背包問題介紹如下: (i)螞蟻演算法 在自然界中之碼蟻行進前往尋找食物時,所行經過之路線會遺留下 落蒙素(ph咖。ne)來辅助找尋食物時,遺留下某魏_外的碼犧 逼其冒經所走過的路徑,隨著越來越多的碼蟻走過相同的路程時,此路 2費落蒙諸縣增加,蚊會絲越高。崎的,其餘魏之路徑費 豕素則會fe.it蒸發域少’最後螞觀可_費錄錄找尋出離食 近的相徑。耻闕理魏是_顯自然界娜行進事 —種人,,進爾—般謝法,達顺尋最輪。第20園為 1284865 然借螞蟻行經之路徑,圖16為所設計之人工碼蟻行經路線。 螞蟻演算法是以旅行推銷員問題(Traveling知化測抑Pr〇b〗⑽,^p) 為求解目標及定義之啟發式演算法,若A為城市i與城市y之間的尤拉距離 β尸收:-不/7价㈣y/刀% 一般會將Tsp的問題以(n,e)來表示,n 、疋斤有城市的集a , E指的是兩兩城市之間的邊界(或稱相連的邊)。且 A(t)(i 1’ ,n)為螞蟻在時間f位於城市』·的數量,故假若所有碼犧的 總數為/» ’ ιη=Σ 4⑴。一開始,隻螞蟻會隨機選擇乃個城市當作起始 點接著碼蟻會隨著機率函數(Pr〇babilistic 進一步選擇碼犧 行走下-個城市據。而機率聽是根㈣洛蒙素(細麵如"⑴ 和局部啟發式疏 W (Locally Heuristic: Functi〇n)。 局。P啟發式函數在本發明中是以兩城市間距離的倒數所取得,故 心片々因此利用下列的方程式表示在時間塒鉍隻螞蟻在城市』到 城市間的機率函數為: pIu): if i e allowed k otherwise 射二之⑴⑽地在㈣所指岐嘯隻_可能聊為下—個城市的 城市允轉合,_與錢控職 、城市之間重要關健度的參數值, 至於費洛象素h之求法為下列方法所求得: rii(t+n)=p>krij(t)+Arij ί P為個係數⑦幼如,相對於我們對費洛蒙素所給的權 至於卜P就是費洛蒙素蒸發的係數值。 31 * 1284865 心細是戦歡城市消物.之間所有費洛蒙素的總和, --為在時間填⑽間間距裡,城市i與吻總共有多少隻碼犧走過,如 此便可利用此定理進行最短逃生路徑之推論。 (i i)背包問題(Knapsack Problem) 利用背包可背負物品之原理定意K的大小將n個不同大小的物品裝 入,每一個大小為ki價值為Vi,使得(ki)並且其總價ν=Σ、(Vi) •最大。其演算法如下所示。(^max — ^min ) ^min ... (11) where I is the independent variable ^ is the average value; r is the standard deviation; 1 and 1 are the minimum and maximum values; ic is the conversion coefficient, when the technique 29 can be 99 · _ data is mapped to [H] interval, and the remaining (four).8, 2.58, 1.96, i 65, h 28 map the variable values of 99 5%, 99%, 9 heat, _, and view to [/ ) — Interval. The lion's inverted transmission neural network data pre-processing analysis method is to obtain the average value and the fresh error of the training machine number for each 26 1284865~ring® sub-, respectively, and A sample is used to convert the variable value domain. (5) Inverted-transfer-like neural network calculation process The operation and calculation process of the inverse-transmission network system can be divided into the following steps: (1) setting the network majority '(2) clearly random random number setting weighting matrix i and [fine , and the partial weight vector ❹ and initial value, (3) calculate the hidden layer (Hidden (four) output; (4) set the output layer 鲁 (Output layer} and the hidden layer allowable gap: (5) calculate the output layer and hidden layer gap; (6) discriminate output Whether the gap between the layer and the hidden layer is larger than the material gap, if the gap is less than the allowable gap 1, the best regression model can be obtained; (7) if the output layer and the hidden layer allow the gap to be larger than the difference, the output layer and the hidden layer weight matrix are calculated. And the bias value correction amount; (8) re-correcting the output layer and the IU layer weight matrix and the partial weight, repeating (3)-(8) until the gap between the output layer and the hidden layer is within the allowable gap (ie, when step (6) is not satisfied), And compare the ambiguity of the sensitivity correction amount to find the most Lujia regression model. The above-mentioned reverse-transfer-like neural network algorithm flow is arranged in the second diagram. (6) Sensitivity analysis of network mode (Sensitivity Analysi s) When using the inverse transfer neural network, if the input and output units are approximately monotonous (H〇m_ous), the sensitivity of the mapping output unit of each unit can be input from the weighted analysis network on the link. A positive positive value indicates a greater positive correlation between the two, and a larger negative value indicates a negative phase between the two. This process represents network learning f and minimizes the energy function with the steepest slope method. Whenever a training paradigm is entered, the network must adjust the weighting value by a large amount of 27.1284865 'small' so that the adjustment amplitude is proportional to the sensitivity of the error function to the weighting value, ie the error function - for the weighted value The partial differential value is proportional to the above, and the above relationship value is defined as the following equation (12): -η 6Ε ~dW~ (12) where the weighting value ^ ^ between the i-th and yth processing units is the learning rate In order to control the error adjustment range of the steepest slope method, the non-immediate factor value and the immediate factor value taken out are finally transmitted back to the back-end expert system for immediate regression operation, so as to correctly predict the potential danger of local soil flow in real time. For this hair According to the value of 181 soil and stone flow risk factors in Nantou County, a database of risk factors for earth-rock flow is established by using the associated database link and object-oriented language. The active disaster notification database system located at the back end of the Internet has its main function. The ECA model is used to actively query and give data, and it is responsible for synchronizing with the WinCE database. In addition, the selection of the escape route is also one of the key points. The functions and methods of the invention proposed and designed mainly include the active database. The architecture and design and escape path simulation, the related design methods are described as follows. (7) Active Database (Active Database) This invention will design and improve the traditional database into a proactive database (Active Database) The ECA (Event-Condition-Action) model is used to process the data in the active database, so that all the information is no longer passively checked. 28 1284865 f2 Stomach material transfer system is the main situation. Check, / / use # $上立立问问衫 to lose people slope collapse information: If there is a new slope, the collapse will be fast Notify the system, let the gamblers query the grammar themselves ((4) 17 Lang Feng), and then open the side of the "sentimental _. In addition, if the user in the vicinity of the area finds that the user is in the danger zone, the system will notify the gambler by means of the police. The inquiry is also made by bribes to achieve the highest effect and the least. Disaster risk. Φ slope collapse damage (4) brother: the construction of the shell library, using single-technology can not solve all the problems, so the use of - architecture for private correlation technology. In recent years, due to the mature development of related technologies such as telemetry technology (10), geographic information system (GIS), global satellite positioning system (GPS) and satellite image processing and analysis system (ips), integrated construction of slope collapse prevention technology and Gps/Gls /Ips technology series application, further active development of slope collapse prevention technology automation and information system, for the subsequent conservation of water and soil resources and prevention of slope collapse disasters, can provide accurate and rapid information exchange, which is beneficial to slope collapse Construction of a disaster environment database operating system. If the application level of the slope collapse disaster prevention technology is applied, the integrated system can be divided into the following three parts: (i) Natural environment: including administrative area, topography, geology, landform, water system, watershed, soil, meteorology, Related data layers such as hydrology and plant distribution. (ii) Humanistic environment, including man-made activities such as land use, road opening, mining, logging, spoil, indiscriminate, indiscriminate burial, and over-construction. (iii) Disaster environment: potential disasters such as landslides, ground slips, earth-rock flows, subsoil seepage, erosion to the source, and soil erosion. 29 1284865 (8) Simulation of escape route When a disaster occurs, the road in the disaster area is damaged, and the victims often appear to be confused and do not know where to escape. The design method of the present invention is called the Adaptive Path Method (APD), which is optimal. The inference of the escape path, and through the virtual reality (VR) simulation of the unrelated route, users only need to use the global satellite positioning system on the pDAM (Gp magic through the wireless network (Mobile Network) to connect to the Internet Decision Support System (DSS) to make the inference of the optimal escape path. The decision support system inference module uses the ant algorithm (eight algorithm) and cooperates with the most secret # algorithm (ShQrtest secret MethQd) to estimate the effective response reality. In the disaster area, the invention uses the backpack problem (Knapsack p(7)_) to make the refuge distance, the occupant capacity, and the special conditions for the survival of the food to ensure that the victims are too concentrated in the refuge, and indeed assist the victims to escape. The ant algorithm and the backpack problem used are introduced as follows: (i) The code of ant algorithm in nature When the ants travel to find food, the route they pass will leave behind the singularity (ph coffee.ne) to assist in the search for food, leaving a certain Wei _ outside the code to force the path that the menstruation has passed, along with As more and more code ants walk through the same distance, this road 2 costs increase in counties, and the mosquitoes will be higher. Saki, the rest of Wei’s path will be fe.it will evaporate less.蚂观可_费录录 Finding the relative path of the food. The shame is Wei _ is the natural world of the Na's marching things - kind of people, Jiner - like Xie Fa, Dashun to find the most round. The 20th garden is 1284865 However, according to the path of ants, Figure 16 shows the artificial code ant route. The ant algorithm is based on the travel salesman problem (Traveling knowledge and measurement PrPb) (10), ^p) Algorithm, if A is the distance between the city i and the city y, the distance of the corpse: - no / 7 price (four) y / knife % will generally be the problem of Tsp expressed as (n, e), n, 疋There is a set of cities a, E refers to the boundary between two cities (or connected sides), and A (t) (i 1 ', n) is the ant located in the city at time f · Quantity, so if the total number of all code sacrifices is /» ' ιη=Σ 4(1). At the beginning, only ants will randomly choose a city as the starting point and then the code ants will follow the probability function (Pr〇babilistic further choose code sacrifice) Walking under the city--according to the probability of the root (four) Luo Mengsu (fine surface such as " (1) and local heuristic W (Locally Heuristic: Functi〇n). P. Heuristic function in the present invention is The reciprocal of the distance between the two cities is obtained, so the heartbeat 々 uses the following equation to express the probability that only the ants in the city to the city at the time: pIu): if ie allowed k otherwise 射二之(1)(10)地在(4) The screaming of the screaming only _ may be discussed as the city of the next city, the value of the important relationship between the _ and the money control, the city, as for the method of the Philo pixel h得: rii(t+n)=p>krij(t)+Arij ί P is a coefficient of 7, as opposed to our weight to pheromone, which is the coefficient value of pheromone evaporation. . 31 * 1284865 The heart is the sum of all the pheromones in the city. - In the interval between time (10), how many cities and kisses have been sacrificed, so you can use this The theorem carries out the inference of the shortest escape path. (ii) Knapsack Problem Using the principle that a backpack can carry an item, the size of K is to be loaded into n different sizes of items, each of which has a value of ki of Vi, such that (ki) and its total price ν = Σ, (Vi) • Maximum. The algorithm is as follows.

Algorithm Knapack(S, K)Input· S(an array of size π storing the sizes of the items),and K(the size if the knapsack) Output: P(a two-dimensional array such that P[i9 k]. exist = true if there exists a solution to the knapsack problem with the first i elements and a knapsack of size k,and P[i,k]· belong =true if ith element belongs to that solution)· Begin P[0, 0]. exist := true; for k :=1 to K do P[0y k]. exist := false; // there is no need to initialize P[i, 0] for i >= L because 32 -1284865 will be computed from P[0, 0] f〇r i := 1 to n do for k := 0 to K do P[i,k]· exist := false; "the default value if P[i-ly k]. exist then P[if k]. exist := true;Algorithm Knapack(S, K)Input·S(an array of size π storing the sizes of the items), and K(the size if the knapsack) Output: P(a two-dimensional array such that P[i9 k]. Exist = true if there exists a solution to the knapsack problem with the first i elements and a knapsack of size k, and P[i,k]· belong =true if ith element belongs to that solution)· Begin P[0, 0 Exist := true; for k :=1 to K do P[0y k]. exist := false; // there is no need to initialize P[i, 0] for i >= L because 32 -1284865 Will be computed from P[0, 0] f〇ri := 1 to n do for k := 0 to K do P[i,k]· exist := false; "the default value if P[i-ly k]. exist then P[if k]. exist := true;

P[i,k]. belong := false; else if k-S[i]>=0 then if P[i-1, k-S[i]]. exist then P[i,k]· exist :=true; P[i9 k]. belong := trueP[i,k]. belong := false; else if kS[i]>=0 then if P[i-1, kS[i]]. exist then P[i,k]· exist :=true; P[i9 k]. belong := true

End 、 ----------—_________ 當判斷災情後,若該使用者位於危險地區則須指引其逃生路境,將逃 生路線展示給予使用者了解並辅助逃生,並利用背包定理選擇適當之避難 斤進行X害之避難。使用者可輸入人數、受傷之情況、老人與孩童等相 關因子進行推論’纽會棚背包定理先將雜所選崎,再·螞蟻定 理將相關之路線描糾來,讓使用者能盡速徹離危險區域並逃生到安全之 避難地。 本I月之種行動化土石流災情預防與通報系統^巾請專利前,既未 曾見諸於働,絲衫開”請前,合伟臟的餅,轉合實用性 及進步性的要件,爰依法提出專射請,歸惠予審查,並賜准專利,實 33 1284865 感德便。 惟以上所述者,僅為本發明之較佳實施例,並非因此而拘限本發明之 專利範圍,舉凡運用本發明專利範圍中所述構造之等效變化,均應包含於 本發明之專利範圍中。End, ----------__________ After judging the disaster, if the user is in a dangerous area, he must guide his escape route, give the escape route display to the user to understand and assist in escape, and use the backpack theorem Choose the appropriate refuge to carry out the evacuation of X harm. The user can input the number of people, the situation of the injury, the related factors such as the old man and the child, and infer that the New Zealand shed backpack theorem first selects the relevant route, and then the ant theorem to trace the relevant route, so that the user can complete the speed Leave the danger zone and escape to a safe haven. This I month's action-based earth-rock disaster prevention and notification system ^ towel before the patent, has not seen in the 働, silk shirt opened "please before, He Wei dirty cake, transfer practical and progressive elements, 爰According to the law, the special scope is requested, and the patent is granted, and the patent is granted. The above is only the preferred embodiment of the present invention, and thus the scope of the patent of the present invention is not limited thereto. Equivalent variations in the construction described in the scope of the present invention should be included in the scope of the invention.

34 1284865 -·· 【圖式簡單說明】 第一圖係本發明主要系統架構圖 第二圖係本發明主要方塊流程圖 第三圖系本㈣行赋制者端_輯目 第四圖係本發明多媒體應用伺服器系統架構圖 第五圖係本發明智慧型代理架構圖 第六圖係本發明智慧型代理流程圖 _第七圖係本發明路徑預測法之空間資料圖 第八圖本發明路徑預測法之屬性資料圖 第九圖本發明路徑預測法之圖形内插法與外插法之推—' 第十圖本發明路徑預測法之路徑預測圖 \ 第十一圖本發明路徑預測法之直向行走圖示 第十二圖本發明路徑預測法之斜向行走圖示 第十三圖本發明土石紐0T衛星影像進行IR與麵统計分析圖 ♦第十四圖本發明專家決策系統架構圖 第十五圖本發明土石流有效集水區分佈劃定示意圖 第十六圖本發明應用gis\rs資訊粹取危險因子示意圖 第十七圖本發明運用RS即時監測土石流危險因子示意圖 第十八圖本發明倒傳遞類神經網路演算流程圖 第十九圖本發明南投縣境内土石流危險因子資料庫 第二十圖本發明自然界螞議行經路線圖 第二十一圖本發明人工螞蟻行經路線圖 35 1284865 - 【主要元件符號說明】 ίο行動式使用者端系統 11基礎座標服務 12客製化服務 13跨異質網路 14即時多媒體傳輸 20多媒體應用伺服器 φ 21智慧型代理 210使用者端 211 網路通訊系統 212使用者介面代理 213監控代理 214雨量資訊代理 215新聞資訊代理 φ 216緊急通報代理 217預先下載代理 218協調代理 219圖層資訊代理 220資料庫 221雨量資訊站與新聞網 22虛擬實境 23 Web化服務 1284865 24 WAP服務 30專家決策系統 31推論引擎 32主動式資料庫 33逃生路徑規劃34 1284865 -·· [Simple description of the diagram] The first diagram is the main system architecture diagram of the present invention. The second diagram is the main block diagram of the present invention. The third diagram is the fourth (four) line of the producer's end _ series of the fourth picture The fifth embodiment of the present invention is a smart agent architecture diagram. The sixth diagram is a smart agent flow chart of the present invention. The seventh diagram is a spatial data diagram of the path prediction method of the present invention. The ninth chart of the prediction method is shown in the figure IX. The path interpolation method and the extrapolation method of the path prediction method of the present invention - 'Tenth diagram The path prediction method of the path prediction method of the present invention\ The eleventh figure The path prediction method of the present invention Thirteenth figure, the oblique path diagram of the path prediction method of the present invention, the thirteenth figure, the IR and surface statistical analysis of the 0T satellite image of the present invention. The fifteenth figure is a schematic diagram of the distribution of the effective water collecting area of the earth-rock flow of the present invention. Figure 16 is a schematic diagram of the application of the gis\rs information extraction risk factor. The present invention uses the RS to immediately monitor the risk factor of the soil flow. Figure 18 is a flowchart of the inverse transfer neural network calculus of the present invention. The nineteenth figure of the present invention is a map of the soil flow risk factor in the Nantou County. Ants route map 35 1284865 - [Main component symbol description] ίο mobile user system 11 basic coordinate service 12 customized service 13 across heterogeneous network 14 instant multimedia transmission 20 multimedia application server φ 21 smart agent 210 use 211 network communication system 212 user interface agent 213 monitoring agent 214 rainfall information agent 215 news information agent φ 216 emergency notification agent 217 pre-download agent 218 coordination agent 219 layer information agent 220 database 221 rainfall information station and news network 22 Virtual Reality 23 Web Service 1284865 24 WAP Service 30 Expert Decision System 31 Inference Engine 32 Active Database 33 Escape Path Planning

Claims (1)

1284865 十、申請專利範圍: 1. 一種行動化土石流災情預防與通報系統,該系統至少包含: 一订動式彳μ麵祕,其縣—終端職,可棘遠_防衫、贿料、多 媒體應用伺服器與專家決策系統中相關資料; 一多媒體應關㈣,至少包含#智慧型代理祕、鮑實齡統及獅及 服務力月b ’以分離巨量之媒體資訊,節省後端之資料庫存取資料之時間,1284865 X. Patent application scope: 1. An action-based earth-rock disaster prevention and notification system, which includes at least: a set-up 彳μ面秘, its county-terminal position, can be far-reaching _ anti-shirt, bribe, multimedia Application server and expert decision-making system related information; a multimedia should be closed (4), at least #Smart agent secret, Bao Shiling system and lion and service power month b 'to separate a huge amount of media information, save back-end data inventory Time of the information, 7仃動式朗麵純及專家決策祕可已最_ _制最新的資訊交換; 、H’4 ,整合行動式制者端純與多媒體應賴服騎提供的資 科’去建置—土石流預測模式,藉以提供使用者進行災前㈣決策判斷之參考。 .如申⑺專利喊第丨項所述之—種行動化土石流災情預防與通報系統, 其仃動式_者端_巾之終端設為—桌上型電腦。 3.— 如申請__丨嫩之—軸化土谢情獅與通報系统 其订動式_者端系財之終端設備係為_筆記型電腦。7 仃 式 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗 朗The prediction mode is used to provide users with reference to the pre-disaster (four) decision-making judgment. Such as Shen (7) patent called the third item - a kind of action-based earth-rock disaster prevention and notification system, the terminal of the squat-style end of the towel is set as a desktop computer. 3.—If applying __丨嫩之—Axis soil Xie lion and notification system The ordering system is the _ notebook computer. =物彳軸丨項靴—種她以峨獅與通報祕 ’、订"使用錢纽中之終職舰為-PDA與智慧型手機。 利峨1雜之—種行输石峨獅與通報系統 6其订如動^財私齡之終職備__手駐行動設備。 其行動式項所述之—種行動化土石峨猶與通報系統 7.如申請專利之終端讀鱗—衛星傳送接收設備。 其行動式物種行動化土石流_防與通報系統 一手持式辦細示; 38 1284865 王球疋位系統接收器提供衛星定位服務或一無線傳輸設備提供無線網路 連結與傳輸; 一電腦攝影機提供多媒體影像擷取; 一麥克風與一耳機提供聲音輸入與輸出; 一手持式設備專用照相機或攝影機; 一將擷取和接收的資料傳送(接收)之顯示器裝置。 8·如申請專利範圍第1項所述之一種行動化土石流災情預防與通報系統, •其行動式使用者端系統之軟體至少包含有:(1)基礎座標服務 (Location-Based Services)、⑵客製化資訊服務(Cust〇mized Information Services)、⑶跨異質網路傳輸(〇ver Heter()gene〇us Networks)與(4)即時多媒體傳輸(Real—timeMultimediaTransmissi〇n)。 9·如申睛專利範圍第1項所述之一種行動化土石流災情預防與通報系統, 其打動式使用者端系統之使用者會策集到之衛星定位資料與土石流災害因 子’包含(1)有效集水區面積、⑵有效溪床長度、⑶有效溪床坡度、 _⑷有麟水區内岩性、⑸有效集水區㈣塌面積、⑹有效累積雨量、 (7)有效降雨強度與⑻植生減八她仙子透過任—設備與行動通訊 網路或網際網路連接上本系統進行通報或推論。 10.如申請專利範圍第i項所述之-種行動化土石流災情預防與通報系 統,其中多媒體應用伺服器之智慧型代理包括有;⑴使用者介面代理、 (2)孤控代理、⑶雨篁資§札代理、⑷新聞資訊代理、⑸緊急通報代 理、⑹預先下載代理、⑺協調代理、⑻圖層資訊代理。 39 1284865 ii.如申請專利範圍第i項所述之一種行動化土石流災情預防與通報系 統,其多媒體應用伺服器裡之多媒體設備至少包含有: 一個人電腦或伺服器提供系統建置與顯示; 一網際網路連線設備提供網路連結與傳輸; 一電腦攝影機提供多媒體影像擷取; 一麥克風和耳機提供聲音輸入與輸出; 一將擷取和接收的資料傳送(接收)之顯示器裝置。 12·如申明專利範圍第丨項所述之一種行動化土石流災情預防與通報系 統,其專家決策系統其組成至少包含有: 一個人電腦或伺服器提供系統建置與顯示; 一網際網路連線設備提供網路連結與傳輸; 一將擷取和接收的資料傳送(接收)之該顯示器裝置。 13·如申請專利範圍第1項所述之一種行動化土石流災情預防與通報系 統,其專家決策系統之軟體元件至少包含有:(U推論引擎(2)主動式資料 庫(3)逃生路徑規劃。 1284865 七、指定代表圖: (一) 本案指定代表圖為:第(二)圖。 (二) 本代表圖之元件符號簡單說明: ίο行動式使用者端系統 20多媒體應用伺服器 30專家決策系統= 彳 彳 丨 — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — Li Wei 1 Miscellaneous - the species of the stone lion and the notification system 6 its order as the mobile end of the private age __ hand stationed mobile equipment. The action-type item is an action-oriented earth-rock 峨 与 通报 通报 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. 7. Its mobile species action-oriented earth-rock flow _ defense and notification system a hand-held office; 38 1284865 Wang ball 疋 system receiver provides satellite positioning service or a wireless transmission device to provide wireless network connection and transmission; a computer camera provides multimedia Image capture; a microphone and a headset provide sound input and output; a handheld device-specific camera or camera; a display device that transmits (receives) captured and received data. 8. A mobile earthquake disaster prevention and notification system as described in item 1 of the patent application scope. • The software of the mobile user system includes at least: (1) Location-Based Services, (2) Cust〇mized Information Services, (3) Trans-heterogeneous network transmission (〇ver Heter()gene〇us Networks) and (4) Real-time Multimedia Transmissi〇n. 9. An action-based earth-rock disaster prevention and notification system as described in item 1 of the scope of the patent application, the satellite positioning data and the earth-rock disaster factor's collected by users of the user-operated user system (1) Effective catchment area, (2) effective bed length, (3) effective creek bed slope, _(4) lithology in the Lianshui area, (5) effective catchment area (4) collapse area, (6) effective cumulative rainfall, (7) effective rainfall intensity and (8) vegetative Decrease her fairy to communicate or infer through the system through the device-operational communication network or the Internet. 10. The mobile-based disaster prevention and notification system as described in item i of the patent application scope, wherein the intelligent agent of the multimedia application server includes; (1) user interface agent, (2) orphan agent, (3) rain § § 代理 agent, (4) news agency, (5) emergency notification agent, (6) pre-download agent, (7) coordination agent, (8) layer information agent. 39 1284865 ii. An action-based earth-rock disaster prevention and notification system as described in claim i, wherein the multimedia device in the multimedia application server comprises at least: a personal computer or server providing system construction and display; The Internet connection device provides network connection and transmission; a computer camera provides multimedia image capture; a microphone and earphone provides sound input and output; and a display device that transmits (receives) the captured and received data. 12. An action-based earth-rock disaster prevention and notification system as described in the third paragraph of the patent scope, the expert decision-making system consists of at least: a person computer or server provides system construction and display; an internet connection The device provides network connection and transmission; the display device that transmits (receives) the captured and received data. 13. For an action-based earth-rock disaster prevention and notification system as described in item 1 of the patent application scope, the software components of the expert decision-making system include at least: (U inference engine (2) active database (3) escape route planning 1284865 VII. Designated representative map: (1) The representative representative figure of this case is: (2). (2) The component symbol of this representative figure is simple: ίο mobile user terminal system 20 multimedia application server 30 expert decision system 八、本案若有化學式時,請揭示最能顯示發明特徵的化學式:8. If there is a chemical formula in this case, please disclose the chemical formula that best shows the characteristics of the invention:
TW94104363A 2005-02-15 2005-02-15 Real-time mobile debris-flow disaster prevention and alert system TWI284865B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW94104363A TWI284865B (en) 2005-02-15 2005-02-15 Real-time mobile debris-flow disaster prevention and alert system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW94104363A TWI284865B (en) 2005-02-15 2005-02-15 Real-time mobile debris-flow disaster prevention and alert system

Publications (2)

Publication Number Publication Date
TW200629183A TW200629183A (en) 2006-08-16
TWI284865B true TWI284865B (en) 2007-08-01

Family

ID=39446005

Family Applications (1)

Application Number Title Priority Date Filing Date
TW94104363A TWI284865B (en) 2005-02-15 2005-02-15 Real-time mobile debris-flow disaster prevention and alert system

Country Status (1)

Country Link
TW (1) TWI284865B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI384423B (en) * 2008-11-26 2013-02-01 Ind Tech Res Inst Alarm method and system based on voice events, and building method on behavior trajectory thereof
TWI406193B (en) * 2009-12-14 2013-08-21 Univ Southern Taiwan Tech Regional mobile alert warning system and method
TWI467077B (en) * 2011-03-17 2015-01-01 Ancad Inc Device and method for making a land subsidence analysis chart
CN104680759A (en) * 2015-02-04 2015-06-03 唐山华洋自动化有限公司 ArcGIS based IC card total pollutant amount monitoring system
TWI580843B (en) * 2014-06-26 2017-05-01 國立中興大學 Conservation methods of sand and sand environment in catchment area

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7898458B2 (en) * 2006-08-03 2011-03-01 Pasco Corporation Disaster countermeasure support method
CN101614822B (en) * 2009-07-17 2011-07-27 北京大学 Method for testing road damage based on post-disaster high-resolution remote sensing image
CN102044127A (en) * 2009-10-14 2011-05-04 中冶(北京)交通科技发展有限公司 Early warning service management system based on remote expert service
CN101826247A (en) * 2010-04-06 2010-09-08 长江水利委员会长江科学院 System for monitoring, forecasting and warning mud-rock flow
JP6941319B2 (en) * 2017-10-10 2021-09-29 日本信号株式会社 Risk assessment system
CN113096360B (en) * 2021-04-21 2022-09-23 贵州大学 Novel analysis monitoring method for slope fixing monitoring

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI384423B (en) * 2008-11-26 2013-02-01 Ind Tech Res Inst Alarm method and system based on voice events, and building method on behavior trajectory thereof
TWI406193B (en) * 2009-12-14 2013-08-21 Univ Southern Taiwan Tech Regional mobile alert warning system and method
TWI467077B (en) * 2011-03-17 2015-01-01 Ancad Inc Device and method for making a land subsidence analysis chart
TWI580843B (en) * 2014-06-26 2017-05-01 國立中興大學 Conservation methods of sand and sand environment in catchment area
CN104680759A (en) * 2015-02-04 2015-06-03 唐山华洋自动化有限公司 ArcGIS based IC card total pollutant amount monitoring system

Also Published As

Publication number Publication date
TW200629183A (en) 2006-08-16

Similar Documents

Publication Publication Date Title
TWI284865B (en) Real-time mobile debris-flow disaster prevention and alert system
Kwan GIS methods in time‐geographic research: Geocomputation and geovisualization of human activity patterns
Chang et al. Combining Google Earth and GIS mapping technologies in a dengue surveillance system for developing countries
Daniel et al. geoSmartCity: geomatics contribution to the smart city
Wood et al. Community variations in population exposure to near-field tsunami hazards as a function of pedestrian travel time to safety
JP5878555B2 (en) View-dependent technology to determine user interest in 3D application features
Taubenböck et al. Risk reduction at the “Last-Mile”: an attempt to turn science into action by the example of Padang, Indonesia
JP6276106B2 (en) Decision support system and decision support method
Sugumaran et al. A web-based environmental decision support system (WEDSS) for environmental planning and watershed management
CN108170714A (en) A kind of three-dimensional simulation system of typhoon disaster monitoring and evaluation
US20140324630A1 (en) System and method for delivering relevant, location-specific gis information to a mobile device
CN110222137A (en) One kind is based on oblique photograph and augmented reality Intelligent campus system
CN103064924A (en) Travel destination situation recommendation method based on geotagged photo excavation
Rafoss et al. Open geospatial technology standards and their potential in plant pest risk management—GPS-enabled mobile phones utilising open geospatial technology standards Web Feature Service Transactions support the fighting of fire blight in Norway
CN106713380A (en) Position sensing based information sharing, pushing and exchanging system
Daniel et al. Watershed Modeling Using GIS Technology: A Critical Review.
Saeidi et al. Assessing the visual impacts of new urban features: Coupling visibility analysis with 3D city modelling
CN106709837A (en) Unapproved construction information high precision acquisition method based on mobile terminal
CN104679942A (en) Construction land bearing efficiency measuring method based on data mining
Alparslan et al. A GIS model for settlement suitability regarding disaster mitigation, a case study in Bolu Turkey
KR20060125169A (en) Spatial information construction technology and disaster mobile gis system
Chen et al. Volunteered geographic information for disaster management with application to earthquake disaster databank & sharing platform
Li et al. Why do we need a national address point database to improve wildfire public safety in the US?
CN109598026A (en) A kind of Landscape Design management system and application method
Zipf et al. Mobile internet GIS based flood warning and information systems