基于支持向量机改进算法的船舶类型识别研究

利用船舶目标辐射噪声DEMON谱特征,采用改进的支持向量机算法,实现了对船舶目标的分类识别研究。针对支持向量机算法对噪声比较敏感和最优分类面求解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两方面进行了改进,提出了基于径向基核函数的齐次决策二阶损失函数支持向量机改进算法,并应用于利用船舶目标辐射噪声DEMON谱进行船舶目标类型分类识别实验。理论分析、数据仿真与实验结果表明,该改进算法实现了在二次规划中的较少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,是一种适合于船舶辐射噪声DENOM分类识别的有效的支持向量机改进算法。

Application of an improved support vector machine classification algorithm to underwater targets recognition

The underwater targets recognition based on DEMON spectra has been realized with a proposed improved support vector machine algorithm. To solve the problem that the support vector machine (SVM) is sensitive to noise and that there are so many restrictions against looking for optimal hyper-plane in solving the optimal hyper-plane, an improvement is done in the relaxation and decision function. The SVM algorithm of homogeneous decision - second order loss function based on RBF (Radial Basis Function) is given, and the solution of optimal hyper-plane under quadratic programming problem is realized. Recognition experiments have been done by using simulation datasets and four kinds DEMON datasets of ship radiated noise. The results show that this algorithm has the characters in large optimized space of model parameters and good overall recognition performance, which is suitable for ship radiated noise recognition.