摘要
该文提出了基于带有偏差单元递归神经网络算法在电容层析成像系统的图像重建中的应用。该算法克服了电容层析成像系统中电容测量灵敏度分布易受被测多相流介质分布的影响,利用MATLAB编程对带有偏差单元的递归神经网络进行训练,给出了图像仿真结果。仿真结果表明,该算法在保证了重建图像质量的前提下,缩短训练时间,提高了成像速度,满足了工业实时性的要求。
This paper proposes a neural network based on IRN with deviation elem ent algorithm and its application in image reconstruction of ECT. The proposed a lgorithm overcomes sensitivity asymmetry influenced by two-phase flow inhomogene ity in ECT. The computer program of IRN with deviation element is drawn up using Matlab. Some simulation results of image reconstruction are given ,which illust rate that the algorithm is effective for image of two-phase flow to be reconstr ucted with accuracy, at a higher speed and can meet the industrial real -time requirement.
出处
《计算机仿真》
CSCD
2005年第4期90-92,96,共4页
Computer Simulation
关键词
两相流
电容层析成像
图像重建
内部回归神经网络
Two-phase flow
Electrical capacitance tomography(EC T)
Image construction
Internal recurrent net(IRN)
