| Summary: | 碩士 === 國立屏東科技大學 === 資訊管理系所 === 96 === In Taiwan, there are usually large-scale debrisflows and landslides in the typhoon season or torrential rain falls. As the rapid progress of wireless sensor networks (WSN) and mobile communication technology, it is the hot issue that using the-state-of-the-art technology to build a reliably disaster prediction model and monitor and collect environment variation information. This paper integrates WSN and Analytic Network Process (ANP) method to evaluate the weight of disaster factors, which is determined using the consistency index of hillslope pairwise comparison. The weight estimation and classly of disaster factors and based on the K-means model to build the hillslope prediction model. Base on the proposed disaster prediction model, we design and implement the “Portrait-based Disaster Alerting System, (PDAS)”. PDAS is four-tier architecture composed of Mobile User Site, Hillslopes Monitoring Sensor Site, Integrated Service Server, and Intelligent Hillslopes Decision System. Mobile User (MU) Site uses the handheld devices to transmit and receive multimedia information about hillslop disasters via the mobile networks. Hillslope Monitoring Sensor (HMS) Site includes the Hillslope Monitoring Engine, network camera, and sensors to collect environment information. Integrated Service Server (ISS) is composed of six intelligent agents to process the received environment information. Intelligent Hillslope Decision System (IHDS) is responsible to determine the probability of hillslope disaster occurrence based on the information from sensors. The hillslope disaster prediction model defines seven disaster factors, which include soil hydrous, rainfall, NDVI, light, displacement, and gradient, to evaluate and analysis the probability of hillslope disaster. The PDAS uses Web-GISto display the environment information visualized. The system evaluation results reveal the proposed prediction model achieves more accuracy disaster determination than the traditional method.
|
|---|
