白石湖吊橋撓度監測在時間序列模式之預測

Research on Time Series Model Prediction for Deflection Monitoring of Baishihu Suspension Bridge

陳智誠(Chih-cheng Chen)；杜敏誠(Min-cheng Tu)；施邦築(Ban-Jwu Shih)；鄭清江(Ching-Jiang Jeng)

撓度 ； 時間序列 ； Deflection ； time series

水保技術

15卷4期（2021 / 12 / 01）

38 - 53

繁體中文

本研究以臺北市內湖區白石湖吊橋中央撓度監測為研究對象，依據過去近10年的每月監測109筆數據，利用其中104筆採用Python的statsmodels時間序列分析與預測，再利用5筆資料驗證，時間序列分解方法包含經典分解（加法與乘法模型）CLASSICAL DECOMPOSITION及STL DECOMPOSITION分解模型作分析，再利用簡單移動平均線（SMA）、加權移動平均線（WMA）、指數移動平均線（EMA）、STL進行預測（EMA）作預測，將分析與預測結果作討論與建議。

In this study, the central deflection monitoring of the Baishihu Suspension Bridge in Neihu District, Taipei City was used as the research object. Based on 109 monthly monitoring data in the past, 104 of them were analyzed and predicted using statsmodels using Python for time series analysis and prediction. The five data was used for verification and time series decomposition. The methods include CLASSICAL DECOMPOSITION classic decomposition (addition and multiplication model) and STL DECOMPOSITION decomposition for analysis, and then use simple moving average (SMA) prediction, weighted moving average (WMA), exponential moving average (EMA), STL for prediction ( EMA) makes predictions, and discusses and recommends the results of analysis and prediction.

1. https://pythonforfinance.net/2019/07/22/time-series-decomposition-prediction-in-python/
2. https://predictivehacks.com/time-series-decomposition/
3. https://zhuanlan.zhihu.com/p/157960700
4. https://www.statsmodels.org/stable/index.html
5. https://www.statsmodels.org/stable/tsa.html
6. https://ithelp.ithome.com.tw/articles/10195400
7. https://otexts.com/fppcn/MA.html
8. https://www.cnblogs.com/cgmcoding/p/14155569.html
9. https://towardsdatascience.com/time-series-decomposition-and-statsmodels-parameters-69e54d035453
10. https://datasciencebeginners.com/2020/11/25/time-series-forecast-and-decomposition-101-guide-python/
11. https://ithelp.ithome.com.tw/articles/10195635
12. 陳智誠,鄭清江,陳世凱,王仲宇(2020)。白石湖吊橋與坡地安全監測及穩定分析案例探討。第18屆大地工程學術研究討論會論文集
13. 智全工程技術顧問有限公司(2020)。,未出版