基于优化核极限学习的WSN网络汇聚节点故障诊断  

WSN Sink Node Fault Diagnosis Based on Optimized Kernel Extreme Learning

作　　者：刘恒 钟俊 刘辉 LIU Heng;ZHONG Jun;LIU Hui(Department of Mechanical and Electrical Engineering,Anhui Vocational and Technical College,Hefei 230011,China)

机构地区：[1]安徽职业技术学院机电工程学院,安徽合肥230011

出　　处：《新乡学院学报》2021年第6期28-32,共5页Journal of Xinxiang University

摘　　要：针对现有无线传感网络sink节点(汇聚节点)故障存在在线诊断效果差、准确率低的问题,提出一种经过优化的核极限学习方法。在核极限学习模型的基础上,从权重和偏置2个层面改善核函数的运算性能;引入了SFLA启发优化算法,提高对sink节点故障分类中全局寻优的性能,增强故障诊断效果。仿真结果表明,所提出的节点故障诊断方法在2种状态下的CPT指标值均趋近于理论值,故障样本的抽样误差率显著优于传统故障诊断方案。Aiming at the poor fault online diagnosis effect and low accuracy of the existing wireless sensor network sink node(sink node),an optimized kernel limit learning method is proposed.On this basis,the operation performance of kernel function is improved from two aspects of weight and bias.The SFLA heuristic optimization algorithm is introduced to enhance the global optimization performance in sink node fault classification and improve the fault diagnosis effect.The simulation results show that the CPT index value of the proposed method approaches to the theoretical value,and the sampling error rate of fault samples is significantly better than that of the traditional fault diagnosis scheme.

关 键 词：优化核极限学习 WSN 核函数 SINK节点 SFLA 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]