RBF神经网络图像重建算法在电容层析成像系统中应用广泛,它较好地克服了ECT系统的软场特性、强非线性和不适定性,其成像时间和成像精确度比其他算法都有很多改善.本文从有限元场域剖分、数据归一化和神经网络输入层角度对该算法进行了相关改进.仿真实验结果证明,改进后的算法有着更好的成像时间和成像精确度.
RBF neural network has higher reconstruction quality and reconstruction speed than other image reconstruction algorithms when used in image reconstruction. It overcame well soft field characteristic, strong nonlinear and ill - posedness of electrical capacitance tomography. This paper improved the algorithm from finite element plotting pattern, data normalization and input layers. The simulation results show the improved method is effective and accurate for image reconstruction.