摘要
为了满足电阻层析成像(electrical resistance tomography, ERT)技术在实际应用时对修正牛顿-拉夫逊算法逆问题求解精度的要求,从初始值、正则化因子调整、灵敏度矩阵更新、阈值优化等方面进行了改进。利用可实现离线优化与计算的改进Landweber预迭代算法获得算法初始值,然后在逆问题求解过程中,按概率采取不同的策略调整正则化因子,同时根据正问题计算结果自动更新灵敏度矩阵,最后以进一步减小敏感场边界电压计算值与测量值的误差为优化目标,利用改进遗传算法优化重建图像阈值,进而得到敏感场域内电阻率分布的最优值。仿真结果验证了上述措施的有效性。
In order to meet the requirements of electrical resistance tomography(ERT) technology for the accuracy of solving the inverse problem of modified Newton Raphson algorithm in practical application, the paper made improvements in the aspects of initial value, regularization factor adjustment, sensitivity matrix update, threshold optimization, etc. Firstly, in the selection of the initial value of the media distribution in the sensitive field, the improved Landweber pre-iterative algorithm was used, which can realize offline optimization and calculation. Then, in the process of solving the inverse problem, different strategies were adopted to adjust the regularization factor according to the probability. Meanwhile, the sensitivity matrix was updated according to the results of the forward problem. Finally, the thresholds of the reconstructed images were optimized to further reduce the error of the values by the improved genetic algorithm, and then the optimal values of the resistivity distribution were obtained. Simulation results verify the effectiveness of the above measures.
作者
周曦
柴晓宇
王彬
肖理庆
ZHOU Xi;CHAI Xiao-yu;WANG Bin;XIAO Li-qing(School of Mechanical and Electrical Engineering,Huainan Normal University,Huainan Anhui 232038,China)
出处
《计算机仿真》
北大核心
2022年第9期252-256,共5页
Computer Simulation
基金
国家级大学生创新创业训练计划项目(202110381003)
安徽省级大学生创新创业训练计划项目(S202110381153)
淮南师范学院第二批校级科研创新团队(XJTD202009)。
关键词
灵敏度矩阵
阈值
概率
Sensitivity matrix
Threshold
Probability
