| Electrical Resistance Tomography(ERT)is a mode of Electrical Tomography(ET),which has the ability to measure the conductivity distribution in a given process equipment.This technology forms a three-dimensional sensitive space under the excitation signal by arranging electrode arrays at equal intervals on the boundary of the measured object field.By measuring the voltage difference between the electrodes,projection data on different observation angles are obtained.Using the appropriate image reconstruction algorithm,the media distribution information in the object field is deduced from the projection data to realize the visualization of the media distribution in the closed container.Due to its non-invasive,non-radiation and visualization characteristics,this technology has become a hot research topic in the fields of multi-phase flow pattern monitoring and medical imaging.In this paper,the three-dimensional finite element model of ERT sensor is established by using finite element simulation software COMSOL.The influence of electrode number,electrode material and electrode shape on sensor performance is analyzed.The influence of electrode size on the sensor performance is emphatically analyzed under the condition of fixing other parameters unchanged.The analysis shows that there is a non-linear relationship between electrode width,length and sensor performance index,which lays a foundation for further research.The sensitive space of sensor is studied in this paper.The sensitive spatial characteristics and axial compressibility of the sensor are explored.Sensitivity d is defined to quantify the sensitivity of the sensor to low conductivity targets located in different spatial positions.The spatial distribution of sensitivity d under different electrode lengths and widths is obtained by finite element calculation.The compression effect is verified by the change of maximum gray value of reconstructed images.The results show that the sensitivity of different spatial positions in the sensor varies greatly.The closer to the axial center of the electrode array and the closer to the tube wall,the higher the sensitivity and the stronger the ability to sense the target.With the increase of electrode width,the sensitive space is compressed inward in the axial direction and the sensitivity difference increases.With the increase of electrode length,the sensitive space expands outward in the axial direction and the sensitivity difference decreases.The axial sensitive space of the sensor can be compressed by increasing the width of the polar plate.In this paper,a sensor optimization method combining response surface method and genetic algorithm is proposed.The optimization parameters are selected as the width and length of electrodes,and the sensitivity matrix condition number,sensitivity field uniformity and sensitivity space area index are integrated into a single optimization objective function.The method firstly adopts a central composite experimental design,obtains response values of sensitivity matrix condition numbers,sensitivity field uniformity and sensitivity space area under different electrode widths and lengths through experiments,and then combines the least square method to carry out response surface analysis according to three groups of response value results under different electrode widths and lengths to obtain three prediction models,and further establishes non-linear relationships among sensor optimization parameters of electrode widths,lengths and sensitivity matrix condition numbers,sensitivity field uniformity and sensitivity space area performance.Three response surface models are built into the iteration of genetic algorithm,and the evolutionary characteristics of genetic algorithm are used to realize the automatic optimization of sensor electrode width and length.The optimization effect of the sensor is verified by synthesizing the objective function and image reconstruction,and both are better improved. |
