摘要
针对漏磁检测中的缺陷反演重构问题,引入了一种新型启发式优化算法—布谷鸟搜索算法,提出了以径向基函数神经网络为前向模型,布谷鸟搜索算法用作迭代算法的漏磁反演方法.为验证该反演方法的有效性,分别使用了不含噪声和含噪声的漏磁仿真信号以及实测漏磁信号.实验结果表明,与粒子群优化算法和差分进化算法相比,布谷鸟搜索算法的处理误差最小,而且对含噪声仿真漏磁信号和实测漏磁信号的重构结果依然能够较好地逼近真实缺陷.因此,基于布谷鸟搜索算法的反演方法对噪声具有一定的鲁棒性,是一种有效可行的漏磁反演方法.
A new heuristic optimization method-cuckoo search (CS) algorithm is introduced and applied to inversing method of magnetic flux leakage (MFL). Radial basis function neural network is regarded as forward model and cuckoo search algorithm is used as iterative algorithm, thus the inversing method is proposed. Simulated MFL signals without noise, with noise and real MFL signals are used to verify the effectiveness of the inversing method, respectively. Experimental results proved the processing error of CS algorithm is smallest by comparing with particle swarm optimization algorithm and differential evolution algorithm. With existence of certain noise in MFL signal, defect profiles reconstructed by the proposed method are still close to the true profiles. The inversing method based on CS algorithm has robustness to the noise and is an efficient reconstructing method. © 2015, Editorial Board of Journal of Basic Science and Engineering. All right reserved.
出处
《应用基础与工程科学学报》
EI
CSCD
北大核心
2015年第6期1275-1283,共9页
Journal of Basic Science and Engineering
基金
国家自然科学基金资助项目(51107080
61304134)
上海市电站自动化技术重点实验室(13DZ2273800)
上海市重点科技攻关计划(14110500700)
关键词
漏磁检测
布谷鸟搜索算法
反演方法
径向基函数神经网络
Algorithms
Evolutionary algorithms
Functions
Heuristic methods
Iterative methods
Learning algorithms
Magnetic flux
Magnetic leakage
Magnetism
Nondestructive examination
Particle swarm optimization (PSO)
Radial basis function networks
