摘要
针对桥梁监测的无线测点优化布置问题,提出一种基于自适应动态惩罚函数的改进广义遗传算法。首先针对无线传感器数量固定和通信距离有限的典型特征将桥梁监测无线测点优化布置表达为约束优化问题,无线传感器的数量和极限传输距离作为优化问题的约束;其次构建了一种能够根据解的偏离程度和种群中高适应度个体数量自动调整惩罚力度的自适应动态惩罚函数;然后采用精英保存机制和末位淘汰策略对基于二重结构编码的广义遗传算法进行了改进;最后利用一大跨悬索桥对该方法进行了验证,并进一步讨论了自适应动态惩罚函数对解的有效性和收敛速度的影响。结果表明:提出的自适应动态惩罚函数能够根据种群的特征自动改变惩罚尺度,保证无线传感器之间的距离小于极限通信距离,同时将无线数据传输距离对桥梁监测信息获取的影响降到最低;改进的广义遗传算法具有很强的全局快速寻优能力,能够快速搜索到全局最优解,优化结果不仅能够满足无线传感网络数据传输距离的要求,还能最大化无线测点的信息获取能力。
An improved generalized genetic algorithm combining with self-adaptive dynamic penalty function was proposed for optimal wireless sensor placement in bridge monitoring.Firstly,the problem of optimal wireless sensor placement in bridge monitoring was modeled as a constrained optimization problem considering the number of wireless sensors and the limited data transmission distance.The number of wireless sensors and the data transmission distance were taken as two constraints.And then,the self-adaptive dynamic penalty function was established,which could adjust the penalty automatically according to the evolution generation and the population of good individuals.Thirdly,the generalized genetic algorithm based on dual-structure coding was improved by elite conservation strategy and worst elimination policy.Finally,numerical experiments were carried out using a long-span suspension bridge and the influences of the self-adaptive dynamic penalty function on the effectiveness of solutions and the convergence speed were discussed.The results indicate that the established self-adaptive dynamic penaltyfunction can minimize the influence of limited wireless data transmission distance on capturing bridge information.The improved generalized genetic algorithm has strong capability of exploring the global optimal solutions and can find the global optimal solution quickly and stably.The optimal wireless sensor configuration extracted by the method proposed can simultaneously meet the data transmission requirement of wireless sensor network and strengthen the capability of obtaining structural information.
作者
周广东
操声浪
刘定坤
ZHOU Guang-dong;CAO Sheng-lang;LIU Ding-kun(College of Civil and Transportation Engineering,Hohai University,Nanjing 210098,Jiangsu,China)
出处
《建筑科学与工程学报》
CAS
北大核心
2018年第5期86-92,共7页
Journal of Architecture and Civil Engineering
基金
国家自然科学基金项目(51678218)
中央高校基本科研业务费专项资金项目(2018B14114)
关键词
结构健康监测
无线传感网络
测点优化布置
广义遗传算法
自适应动态惩罚函数
structural health monitoring
wireless sensor network
optimal sensor placement
generalized genetic algorithm
self-adaptive dynamic penalty function