Building The Rainfall-Runoff System By Knowledge Engineering

• Main Authors: Jia-wei Liang, 梁家瑋
• Other Authors: Chang-sian Chen
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/34765163504260273716

碩士 === 逢甲大學 === 水利工程所 === 95 === Because the terrain of Taiwan is precipitous, the rain tendency is centralized and the river flows urgently short. The regional rainfall that makes is influenced by topography greatly. The rainfall is combined by many kinds of factors. It also influence by time and space. For the complicated and nonlinearity system, statistics method and physics method both are difficult to effective treatment. Carry on the way to predict to the hydrological data after presenting many kinds of artificial intelligence field. But because a great characteristic of artificial intelligence, so long as can solve the problem, with the accumulative total of time and materials, every way can reach the good result. How to reduce resources and waste of time is the way to compare quality. This research tries to combine artificial intelligence with knowledge engineering. According to six major stages of knowledge engineering, design one artificial intelligence system and construct the prediction model in rainfall-runoff. Build the prediction way in which by the artificial intelligence system. BPN comes to deal with to nonlinearity of the relation. Knowledge system cooperate with Euclidean distance conduct comparing with the time of dealing with the way. And try to revise the uncertainty of solving with the fuzzy theory. With validation tests at Wu-Xi watershed, the models get good prove of results. In two kinds of common methods used treatment rainfall. It also can get good result without change models’ structure.