Traffic Flow Data Stitching for Information Service-a Fuzzy Time Series Approach

• Main Author: 陳哲先
• Other Authors: 艾嘉銘
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/77274455616785106010

碩士 === 逢甲大學 === 運輸科技與管理學系 === 102 === Past studies, most of the fuzzy time series (Fuzzy Time Series) is applied to the commercial port cargo forecasts or information is more stable relative data to predict the trend in some, there is no research data used in large and dramatic changes in the amount of traffic over the information on. In this study, the application of the ARIMA (1,1,1) model, the highway vehicle detectors (Vehicle detector) data input after the predicted value and then get into the ARIMA time series fuzzy arithmetic prediction, detection methods such mixed more direct data input fuzzy time series forecasting mean absolute error (MAPE) of 35.67 down to 10.64, as can be inferred if the fuzzy time series and then corrected to fit the data used to fill vehicle detectors, with the filter function can have algorithm should be able to improve model accuracy and because of the time series of fuzzy logic algorithms are more simple, fast processing speed, simplify the algorithm to fill the current model is too complex a problem, refer to the highway managers as a way of actual applications.