頁籤選單縮合
題 名 | 交通事件偵測之新方法--混沌異常車流診斷法=Chaos Abnormal Traffic Diagnosis for Traffic Incident Detection: A New Approach |
---|---|
作 者 | 藍武王; 林豐裕; 黃業傑; | 書刊名 | 交通運輸 |
卷 期 | 22 2003.06[民92.06] |
頁 次 | 頁51-68 |
分類號 | 557.34 |
關鍵詞 | 交通事件偵測; 異常車流診斷; 混沌交通參數; Abnormality traffic diagnosis; Traffic incident detection; Chaotic traffic parameters; |
語 文 | 中文(Chinese) |
中文摘要 | 本文提出一種新的交通事件偵測原理—混沌異常車流診斷法。該方法基本原理係利用混沌交通參數的改變,包含最大里亞譜諾夫指數(largest Lyapunov Exponent)、相關維度(correlation dimension)、相對複雜度(relative Lz complexity )、相關時間(correlation time)和Hurst指數等指標,來判斷交通事件是否發生。經中山高速公路之實證顯示,交通事件發生前後最具明顯變化的指標為最大里亞譜諾夫指數,故本文以其作為車流異常診斷指標。進一步以Paramics模擬中山高速公路主線雙車道上不同之交通載情境,觀測事件前後最大里亞譜夫指數(λ)之變化情形,結果發現當車流正常時λ大於0.49,當車流異常時λ小於0.49,因此以最大里亞譜諾夫指數λ=0.49作為判斷事件是否發生之門檻值。經離線測試顯示,此一門檻值來偵測交通事件之平均偵測正確率高達93.75%,平均誤報率則為2.6%。 |
英文摘要 | This paper proposes a new approach for traffic incident detection—chaos abnormal traffic diagnosis. The underlying theory for this new approach is to measure the change in chaotic traffic parameters, including largest Lyapunov exponent, correlation dimension, relative Lz complexity, correlation time, and Hurst exponent, to examine the existence of traffic incidents. The empirical evidence on the Freeway No.1 shows that the chaotic traffic parameter with most significant change before and after a real traffic incident case is the largest Lyapunov exponent; thus it is selected as an index for the traffic abnormality diagnosis. The traffic simulator—Paramics is further used for simulating and examining the range of the largest Lyapunov exponent (λ) under different traffic incident scenarios on a two-lane freeway mainline. It is found that under the normal traffic conditionsλ is greater than 0.49; under the incident traffic conditionsλis less than 0.49. Thus, λ=0.49 is used as the threshold value for distinguishing the abnormal traffic from the normal one. The off-line tests conclude that the overall average detection rate can reach 93.75% with an average false alarm rate of 2.6%. |
本系統中英文摘要資訊取自各篇刊載內容。