基于改进联合稀疏电容层析成像算法滑油监测研究

Lubricating Oil Monitoring Based on Improved Joint Sparse Electrical Capacitance Tomography Algorithm

关键词:  电容层析成像  滑油监测  图像重建  联合稀疏  对称交替方向乘子法

Key words:  Electrical capacitance tomography  Lubricating oil monitoring  Image reconstruction  Joint sparse  Symmetric alternating direction multiplier method

摘要:

由于电容层析成像（Electrical Capacitance Tomography，ECT）的“软场”特性，使其应用于航空发动机滑油管道监测时难以准确获得被测对象的位置和大小信息。针对图像重建过程中的病态问题，在对称交替乘子法（S-ADMM）算法基础上，提出一种采用L2,1-范数作为核范数、(L1-εL2)范数作为正则化器的改进联合稀疏电容层析成像算法。仿真实验采用COMSOL5.3和MATLAB2014a搭建仿真实验平台，仿真结果表明，改进算法与Landweber迭代算法相比，成像误差降低了35.82%，相关系数提高了56.51%；与S-ADMM算法相比成像误差降低了35.02%，相关系数提高了41.82%，图像误差降低至0.1435、相关系数提升至0.9308，且成像时间保持在0.1s。实验结果表明，改进联合稀疏成像算法提高了滑油监测方面的成像质量，具有实效性和适用性。

Abstract:

Due to the ‘soft field’ characteristic of electrical capacitance-tomography （ECT）， it is difficult to accurately obtain the position and size information of the measured object when it is applied to the monitoring of oil pipeline in aeroengine. Aiming at the ill-posed problem in image reconstruction， an improved joint sparse electrical capacitance tomography algorithm using L2,1norm as kernel norm and (L1-εL2)norm as regularization is proposed based on S-ADMM algorithm. In the simulation， COMSOL5.3 and MATLAB2014a were used to build the simulation platform. The simulation results show that compared with Landweber algorithm， the imaging error of the improved algorithm was reduced by 35.82%， the correlation coefficient increased by 56.51%. Compared with S-ADMM algorithm， the imaging error of the improved algorithm was reduced by 35.02%， the correlation coefficient increased by 41.82%， the image error was reduced to 0.1435， the correlation coefficient increased to 0.9308， and the imaging time was maintained at 0.1s. The experimental results show that the improved algorithm improves the imaging quality of lubricating oil monitoring and has the effectiveness and applicability.