Research Of PXI-based Three-dimensional Multi-frequency Electrical Impedance Tomography System

Electrical impedance tomography (Electrical Impedance Tomography EIT) technology is a new measurement technique in recent years due to its advantages without radiation, non intrusion, and simple structure in low cost. In the clinical monitoring and industrial measurement, it has broad applications and anticipated prospects. In the last twenty years,although the electrical impedance tomography technology has made great progress, but due to the severe ill-pose to inverse problem solution, the reconstructed image quality is not as ideal as expected. Most of the researchers have paid more attention to the research on image reconstruction algorithm, while the design of hardware and software is relatively less and singularity.In this thesis, the neural network technology is studied to establish a neural network EIT model, to optimize the collected data, and to improve the quality of the reconstructed image. A multi-frequency PXI-based electrical impedance tomography system is proposed and completed in the research.The experimentations are undertaken in the physical experiment sink to realize three-dimensional static and dynamic electrical impedance tomography. The EIT data collected at different frequencies are fused through trained neural network to improve the quality of the reconstructed image. The main research work is as following:1. Method research for improving the quality of EIT images by using neural network.Based on the principle of electrical impedance tomography, a three layer neural network of 208-10-208 is constructed. The measured data collected on the experimental platform are divided into training data set and imaging data set, and the simulation data is used as the expected value. First, take the training data as the input of the neural network, the neural network was trained to obtain neural network parameters, the neural network model is established, and then take the imaging data as the neural network input ,using output data from neural network to reconstruct image, finally evaluated the image using six objective indexes.2. A set of multi-frequency electrical impedance tomography system based on PXI bus is built. This system takes the PXI cards made by NI company as the hardware foundation,uses the LabVIEW software to compile the control program and realize the data acquisition,finally the collected data is uploaded to the PC and the image reconstruction algorithms are run by Matlab programms. The control functions of the PXI-based EIT system mainly includes: the setting of the frequency and amplitude of the excitation signal, there are four different frequencies options as 100KHz. 200 KHz, 400 KHz. 800KHz. and the amplitude is set to be 5mA. The switching board can switch working patterns between 2D and 3D models. The two-dimensional model use adjacent excitation and adjacent measurement while 3D model use the quasi diagonal excitation in diffirent layers and the adjacent measurement in the same layer. The sampling frequency of data also can be selected from 5MHz, 10MHz, 20MHz, and 40MHz.3. A kit of software progeammed by Lab VIEW was used to realize the digital demodulate by seveal sub-VI functions. The software can also acquisite the voltage data for digital demodulation, get the real and imaginary parts of the impedance information(impedance information), calculate the amplitude and reconstruct images.4. The test paltform with two layers of 16 electrodes (8 electrodes seperatively) was manufactoried whose simulation physical model was established by COMSOL software.The sensitivity matrix of the physical model is calculated, and the 3D simulation images are reconstructed by the conjugate gradient algorithm.5. The experimentations are carried out by the developed hardware and software kit for the PXI-based 3D EIT system. The static and dynamic images are reconstructed on the experimental water tank by using the non identical quasi diagonal excitation of the same layer. The imaging results can reflect the shape, position and motion state of the target object satisfactorily in the test tank.6. The data fusion method is studied using Lab VIEW and MATLAB software. The method is objective to optimize the voltage data measured at different frequencies using data fusion technology, and to improve the data accuracy and reliability for good quality of the reconstructed images.The innovation of this research work is to realize a PXI-based 3D EIT multi-frequency data acquisition system, and to use neural network and data fusion methods to optimize the collected data and improve the image quality of reconstructed images.