支持向量机改进算法在船舶桨叶数分类中的应用

Application of an improved support vector machine classification algorithm for ship propeller blade number recognition

利用船舶目标辐射噪声DEMON谱特征,采用支持向量机改进算法,实现了对船舶螺旋桨桨叶数的分类识别应用研究。针对支持向量机算法对噪声比较敏感和求解最优分类面时约束较多不利于支持向量机最优分类面寻优的问题,在保持支持向量稀疏特性和应用径向基核函数的条件下,对支持向量机算法在松弛变量和决策函数2方面进行改进,构造齐次决策二阶损失函数径向基支持向量机改进算法,进行理论分析、数据仿真实验,并应用于利用船舶目标辐射噪声DEMON谱进行船舶螺旋桨桨叶数的分类识别实验。结果表明,该改进算法实现了支持向量机在二次规划中的最小约束条件下最优分类面求解,具有模型参数寻优空间广阔、总体分类性能优的特点,其分类性能优于原支持向量机算法,适用于利用船舶目标辐射噪声DENOM进行船舶螺旋桨桨叶数分类应用。

Realized ship propeller blade number recognition base on DEMON with a proposed improved support vector machine algorithm. To solve the problem that support vector machine was sensitive to noise and there were so many restrictions in solving optimal hyperplane,which made against looking for optimal hyperplanea,an improvement was done in relaxation and decision function. Homogeneous decisionsecond order loss function RBF SVM was given,and the solution of optimal hyperplane under quadratic programming problem was realized. Recognition experiments were done using simulation datasets and five kinds DEMON datasets of ship radiated noise. The results showed that this algorithm had the characters that the model parameters optimized space was larger and the overall recognition performance was good,which was suitable for ship propeller blade number recognition.