常婧 柳曉鳴 李夢蕊



摘要:為對船舶航行安全狀況進行有效預測,利用支持向量回歸(support vector regression, SVR)算法構(gòu)建船舶航行安全評估模型。在分析影響船舶航行安全的因素的基礎上,對船舶歷史數(shù)據(jù)進行預處理后將其作為模型訓練和測試的樣本數(shù)據(jù)。實驗結(jié)果表明:該模型評估準確度可達99.6%以上;在同一樣本數(shù)據(jù)條件下,模型的評估準確度和穩(wěn)定性均優(yōu)于基于極限學習機(extreme learning machine,ELM)構(gòu)建的模型。模型的評估結(jié)果為水上交通管理部門的監(jiān)管提供參考。
關鍵詞: 航行安全; 安全評估; 支持向量回歸(SVR); 極限學習機(ELM)
中圖分類號: U676.1 ? ?文獻標志碼: A
Ship navigation safety assessment model based on SVR
CHANG Jing, LIU Xiaoming, LI Mengrui
(Information Science and Technology College, Dalian Maritime University, Dalian 116026, Liaoning, China)
Abstract: In order to effectively predict the ship navigation safety status, the ship navigation safety assessment model is constructed by the support vector regression (SVR) algorithm. Based on the analysis of the factors affecting ship navigation safety, the ship historical data are preprocessed and used as sample data for model training and testing. The experimental results show that: the accuracy of the model evaluation can reach more than 99.6%; under the same sample data conditions, the accuracy and stability of the model are better than those constructed based on the extreme learning machine (ELM). The assessment results of the model provide reference for the supervision of the water traffic management department.
Key words: navigation safety; safety assessment; support vector regression (SVR); extreme learning machine (ELM)
0 引 言
水上交通運輸安全已成為制約我國航運業(yè)快速發(fā)展的重要因素。船舶在航行過程中受到人員、環(huán)境、船舶以及管理等多方面因素的影響,對船舶的航行安全狀況進行評估需要全面考慮這些因素。結(jié)合與這些因素相關聯(lián)的歷史事故數(shù)據(jù),構(gòu)建一個航行安全評估模型,實現(xiàn)對航行安全狀況的有效評估是當前急需解決的問題。
目前,評估船舶航行安全的方法主要有層次分析法(analytic hierarchy process,AHP)、主成分分析(principal component analysis,PCA)法、灰色理論、綜合安全評估(formal safety assessment,F(xiàn)SA)、模糊綜合評價(fuzzy comprehensive evaluation,F(xiàn)CE)以及人工神經(jīng)網(wǎng)絡(artificial neural network,ANN)等方法。張笛等[1]從靜態(tài)信息、動態(tài)信息、通航環(huán)境和自然環(huán)境等方面對船舶數(shù)據(jù)信息進行了分析,利用AHP對內(nèi)河船舶航行安全狀態(tài)進行了評價。李款等[2]基于橫傾角、舵角、艇速3個指標數(shù)據(jù)信息,分別利用PCA和BP神經(jīng)網(wǎng)絡方法對水面艇的航行狀態(tài)進行評估。賈明明等[3]基于自然環(huán)境、航道環(huán)境和交通環(huán)境3類指標及其權(quán)重,利用灰色模糊綜合評價方法量化評估通航安全等級。……