摘要
大型结构由于传感器数量庞大、种类众多,使得对结构进行损伤识别时的速度和准确性不高。针对该问题,设计一种改进型免疫算法。将算法的抗体种群划分为多个子种群并行搜索,减小种群的搜索空间;采用阶段式阈值方式选取算法中的记忆库抗体,使得记忆库随着抗体性能不断调整;利用改进后的算法对各类传感器数据分别进行损伤识别;采用D-S证据理论将位移、应力、加速度多类传感器的识别结果进行数据融合,消除多传感器数据间的不确定因素。实验结果表明,改进型免疫算法能够以较少的迭代次数寻找到最优解,从而提高识别速度,D-S证据融合多类传感器的识别结果后准确度更高。
Due to the large number and variety of sensors,the speed and accuracy of damage identification for large structures are not high.To solve this problem,we designed an improved immune algorithm.The antibody populations of the algorithm were divided into several sub-populations and searched in parallel to reduce the searching space of the population.We adopted the stage threshold method to select the memory library antibody in the algorithm,which made the memory library adjust continuously with the performance of the antibody.The improved algorithm was used to identify the damage of various sensor data separately.Then we used D-S evidence theory to fuse the identification results of displacement,stress and acceleration sensors,and eliminated the uncertainties among the multi-sensor data.The experimental results show that the improved immune algorithm can find the optimal solution with less iteration,which improves the identification speed.The D-S evidence fusion improves the accuracy of multi-sensor identification results.
作者
曹豪
俞阿龙
季佳佳
Cao Hao;Yu Along;Ji Jiajia(College of Electrical Engineering and Control Science,Nanjing Tech University,Nanjing 211800,Jiangsu,China;School of Physical and Electronic and Electrical Engineering,Huaiyin Normal University,Huaian 223300,Jiangsu,China)
出处
《计算机应用与软件》
北大核心
2019年第5期311-315,共5页
Computer Applications and Software
基金
江苏省高校自然科学重大研究项目(16KJA460003)
关键词
大型结构
改进免疫算法
D-S证据理论
多传感器数据融合
损伤识别
Large structure
Improved immune algorithm
D-S evidence theory
Multi-sensor data fusion
Damage identification
