Evaluation of Critical Groundwater Level in Landslide Area

• Main Authors: Bo -Wen Chang, 張博雯
• Other Authors: Miau-Bin Su
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/41274912779604610623

碩士 === 國立中興大學 === 土木工程學系 === 88 === Landslide which occurred at Li shan in April 1990 caused road service interrupted and some building in this area cracked and subsided. Infiltration of rain water via percolation from surface and wild crack is concluded as the main factor influencing the stability of slopeland. According to detailed investigation, experts prove the landslide is active currently. A predefined procedure for the judgement of landslide occurrence has to be set in advance.The goal of this research is to implement an early warning system and to mitigate disaster due to sudden collapse of the slide. The relationship between rainfall intensity and groundwater level fluctuation is learned by artifical neural network . For efficient learning, the choices about neurons of input layer are decided by transfer function. Based on the proposed neural network model, the groundwater level fluctuation responses the difference in its hydrogeology between the working period of drainage facilities and the drained validity of drainage as hourly data on rainfall is given. Furthermore, the detailed geotechnical investigations and the results of the field monitoring system support the explanation of the critical geological layers inside slope. At last, a method to evaluate critical groundwater level with artificial neural network is proposed to form the basis of slopeland management in landslide areas.