论文部分内容阅读
电容层析成像系统在工业上用于管道内流体的实时监测，本文提出用神经网络的方法来代替传统的反投影算法进行图像重建，以解决传统算法存在的扭曲形变和空间精度受限的问题．文中用改进的ＢＰ算法训练，采用一个两层感知机网络，网络的输入是预处理过的测量电容矢量，输出直接对应到空间图像．实验结果表明使用这一方法，能重建出足够精确的图像
Capacitance tomography system is used in industry for real-time monitoring of fluid in pipeline. In this paper, neural network method is proposed instead of traditional back-projection algorithm to reconstruct image to solve the problems of traditional algorithm such as distortion distortion and limited spatial precision . In this paper, an improved BP algorithm is used to train a two-layer sensor network. The input of the network is a preprocessed measurement capacitance vector, and the output corresponds directly to the space image. Experimental results show that this method can reconstruct a sufficiently accurate image