应用声学
應用聲學
응용성학
APPLIED ACOUSTICS
2010年
3期
206-211
,共6页
舰船辐射噪声%支持向量机%径向基核函数%分类
艦船輻射譟聲%支持嚮量機%徑嚮基覈函數%分類
함선복사조성%지지향량궤%경향기핵함수%분류
Ship-radiated noise%Support vector machine%Radial basis function%Classification
本文采用径向基核函数的支持向量机的分类算法,实现了对舰船目标的分类识别.对两类不同类型的舰船的辐射噪声的DENOM谱建立了支持向量机模型,并进行了分类识别试验.试验结果表明,在结构风险最小的准则下,采用网格搜索法确定,径向基核函数的参数σ取值0.23、惩罚系数C值取13为最优的分类识别参数.并通过留一法验证,该模型具备良好的推广能力,总体正确识别率为91.2%.
本文採用徑嚮基覈函數的支持嚮量機的分類算法,實現瞭對艦船目標的分類識彆.對兩類不同類型的艦船的輻射譟聲的DENOM譜建立瞭支持嚮量機模型,併進行瞭分類識彆試驗.試驗結果錶明,在結構風險最小的準則下,採用網格搜索法確定,徑嚮基覈函數的參數σ取值0.23、懲罰繫數C值取13為最優的分類識彆參數.併通過留一法驗證,該模型具備良好的推廣能力,總體正確識彆率為91.2%.
본문채용경향기핵함수적지지향량궤적분류산법,실현료대함선목표적분류식별.대량류불동류형적함선적복사조성적DENOM보건립료지지향량궤모형,병진행료분류식별시험.시험결과표명,재결구풍험최소적준칙하,채용망격수색법학정,경향기핵함수적삼수σ취치0.23、징벌계수C치취13위최우적분류식별삼수.병통과류일법험증,해모형구비량호적추엄능력,총체정학식별솔위91.2%.
In this paper, adoption of support vector machine with radial basis function kernel classification algorithm, succeed in realizing ship targets classification.Establish support vector machine models to two different rypies of ship-radiated noises DEMON spectrum, and the classified recognition experiment has been done.The experimental result indicates that, under the standard of structural risk minimization and adopting grid-search method, the radial basis function kernel parameter σ value 0.23 and the penalty parameter C value 13 are the most superior classification parameter.Meanwhile, this model has good capability in generalizing according to the validating by "leave-one-out" method, and the total correct identification probability is 91.2%.
