摘要
为解决舰船无线通信网的“小样本-高噪声”问题,避免过拟合,研究大数据驱动下舰船无线通信网异常状态辨别方法。采集舰船无线通信网大数据,提取舰船无线通信网状态特征,通过大数据驱动的半监督学习算法为未知标签的无线通信网状态特征样本生成高可信度的伪标签,以已知标签样本和带伪标签的未知样本为大数据驱动的最小二乘半监督支持向量机模型的输入,输出舰船无线通信网异常状态辨别结果。实验证明,该方法可有效采集舰船无线通信网大数据,并提取网络状态特征;该方法异常状态辨别残差的最高自相关系数约为0.2,即辨别结果与实际结果的差距小,异常状态辨别精度高。
To address the"small sample-high noise"issue in ship wireless communication networks and avoid overfitting,a method for identifying abnormal states in ship wireless communication networks driven by big data is studied.Big data from ship wireless communication networks is collected,and state features of the networks are extracted.A semi-supervised learning algorithm driven by big data is used to generate high-confidence pseudo-labels for wireless communication network state feature samples with unknown labels.Known label samples and unknown samples with pseudo-labels are input into a least squares semi-supervised support vector machine model driven by big data,and the output is the identification result of abnormal states in ship wireless communication networks.Experiments demonstrate that this method can effectively collect big data from ship wireless communication networks and extract network state features.The highest autocorrelation coefficient of the residuals in abnormal state identification is approximately 0.2,indicating that the difference between the identification result and the actual result is small,and the accuracy of abnormal state identification is high.
作者
仇丹丹
段新华
QIU Dandan;DUAN Xinhua(Puyang Institute of Technology,Henan University,Puyang 457000,China;Puyang Vocation Technology College,Puyang 457000,China)
出处
《舰船科学技术》
北大核心
2025年第16期185-189,共5页
Ship Science and Technology
关键词
大数据驱动
舰船通信网
异常状态辨别
支持向量机
data-driven
ship communication network
abnormal state identification
support vector machine
作者简介
仇丹丹(1984-),女,硕士,副教授,研究方向为人工智能、云计算及大数据技术。
