| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊工程學系 | zh_TW |
| DC.creator | 翁正剛 | zh_TW |
| DC.creator | Jheng-Gang Wong | en_US |
| dc.date.accessioned | 2020-7-28T07:39:07Z | |
| dc.date.available | 2020-7-28T07:39:07Z | |
| dc.date.issued | 2020 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=107522121 | |
| dc.contributor.department | 資訊工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 地下水資源為台灣重要的用水來源，近年來由於台灣濁水溪沖積扇區域的嚴重超抽地下水，造成濁水溪沖積扇的地層下陷，因此預測地下水位變化是很重要的。有許多關於地下水資源的研究，但這些研究所使用的資料都沒有實際地下水抽取量的資料，有研究指出真實的地下水抽取資料，可以用抽地下水時所會用到的抽水馬達的用電量，來推估地下水抽取量。本研究以線性回歸(Linear Regression)、支持向量機(Support Vector Machine)、隨機森林(Random Forest)和人工神經網路(Artificial Neural Network)等不同的機器學習演算法，建構預測台灣濁水溪沖積扇區域地下水位變化之模型；本研究的特徵資料為約16萬多個分布於濁水溪沖積扇區域的電錶用電資料，以此資料進行機器學習演算法的建模，預測濁水溪沖積扇區域之第一和第二含水層的39個觀測井未來的地下水位。找取每口觀測井之最佳資料貢獻半徑，以提升模型表現。由於第一含水層與第二含水層的地下水變化之相關性低，所以將屬於第一含水層和第二含水層的觀測井個別分開進行實驗
。結果表示對於第一和第二含水層的未來30天每一天平均地下水位之預測各別可達MSE為0.998, 1.508、R^2為0.995, 0.996。 | zh_TW |
| dc.description.abstract | In recent years, due to the severe over-extraction of groundwater in the area of the alluvial fan of the Turbidity Creek in Taiwan, the formation of the alluvial fan of the Turbidity Creek has subsided. Therefore, it is important to predict the change of the groundwater level. There are many studies on groundwater resources, but the data used in these researches does not have data on actual groundwater extraction. Some studies have pointed out that real groundwater extraction data can use the power consumption of the pumping motor that will be used when pumping groundwater. To estimate the groundwater extraction. This research uses different machine learning algorithms such as Linear Regression, Support Vector Machine, Random Forest and Artificial Neural Network. Construct a model to predict the change of groundwater level in the area of the alluvial fan of Choushui River in Taiwan. The characteristic data of this study are more than 160,000 meters of electricity consumption data distributed in the area of the alluvial fan of Choushui River, which is used to model the machine learning algorithm to predict the first and second aquifers in the area of the alluvial fan of Choushui River. Of the 39 observation wells of the future groundwater level. Find the best influence radius of each observation well to improve the model performance. Since the correlation between groundwater changes in the first aquifer and the second aquifer is low, the observation wells belonging to the first aquifer and the second aquifer are separately modeled. The results indicate that the average groundwater level for each of the first and second aquifers in the next 30 days can be predicted to reach MSE of 0.998 and, 1.508, respectively. The results of R^2 are 0.995 and, 0.996, respectively. | en_US |
| DC.subject | 地下水位 | zh_TW |
| DC.subject | 濁水溪沖積扇 | zh_TW |
| DC.subject | 機器學習 | zh_TW |
| DC.subject | 類神經網路 | zh_TW |
| DC.subject | 支持向量機 | zh_TW |
| DC.subject | 線性回歸 | zh_TW |
| DC.subject | 隨機森林 | zh_TW |
| DC.subject | Groundwater | en_US |
| DC.subject | Choushui River Alluvial Fan | en_US |
| DC.subject | Machine Learning | en_US |
| DC.subject | Artificial Neural Network | en_US |
| DC.subject | Support Vector Machine | en_US |
| DC.subject | Linear Rergression | en_US |
| DC.subject | Random Forest | en_US |
| DC.title | 利用機器學習預測濁水溪沖積扇區域之地下水位 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Using Machine Learning to Predict Groundwater Level in Choushui River Alluvial Fan Area | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |