Tin Dioxide-based Gas Sensor Array For Transformer Fault Characteristic Gas Detection And Its Characteristics

Dissolved gas analysis?DGA?is an important means of condition monitoring of oil-immersed power transformer.Accurate and timely acquisition of characteristic gas components and concentration of transformer faults is the key to DGA of online monitoring.As the core of dissolved gas analysis in transformer oil,gas sensing technology will directly affect the validity and accuracy of detection results.Gas sensor array technology is an important part of state monitoring and intelligent sensing,which can simplify the complex structure and detection process of transformer fault characteristic gas monitoring.SnO₂-based gas sensor is a common sensor for main dissolved fault characteristic gases?H₂,C₂H₂,CO,etc.?in transformer oil,and metal doping can significantly improve its detection characteristics.At present,the mechanism of metal doping modification of SnO₂-based gas sensing materials is not clear yet,and there is few related research on the detection of fault characteristics gases in transformer oil based on gas sensor array.In this paper,the H₂,CO and C₂H₂ were detected as target gases,the modified gas sensing mechanism of metal-doped SnO₂-based sensitive materials is systematically studied based on the first principles;gas sensor array is developed by combining finite element simulation,hydrothermal synthesis method and micro-electromechanical machining;the detection characteristics to dissolved gases in transformer oil of gas sensor array are tested,and the gas sensing mechanism of metal doping modification was verified and perfected;Qualitative identification and quantitative estimation of mixed gases by gas sensor array are studied.The research results have important theoretical and practical significance for improving the detection level of transformer fault characteristic gases.The main research work of this paper is as follows:?1?The mechanism of SnO₂-based metal doping modification for gas sensing is systematically studied,which provides guidance for the selection of sensitive materials for gas sensor array.The doping models of pure and 13 common metal-doped Sn₁₂O₂4?110?surfaces such as silver?Ag?,gold?Au?,cobalt?Co?,chromium?Cr?,copper?Cu?,indium?In?,molybdenum?Mo?,nickel?Ni?,palladium?Pd?,platinum?Pt?,titanium?Ti?,tungsten?W?,zinc?Zn?and their gas adsorption models for H₂,CO and C₂H₂ were established,and the micro-parameters such as doping formation energy,gas adsorption energy,atomic configuration,energy band structure,differential charge density,electronic state density and charge distribution were calculated and analyzed.It was found that the d-electron orbital of doped metal atoms strongly hybridized with SnO₂ system to form a new peak of state density,which reduced the band gap width of the doped system.The charge transfer amount of H₂ adsorbed on Ag and Cu doped SnO₂ surface is the largest?0.20 e?,Pd doped system is the largest?0.25 e?,and C₂H₂ is the largest?0.26 e?on Ti and Zn doped system.Considering the band gap width,gas adsorption energy and transfer charge,the sensitive materials of gas sensor array are pure SnO₂ and metal Ag,Au,Cu,Mo,Pd,Pt and Zn doped SnO₂,respectively.?2?The simulation design and development of metal-doped SnO₂-based gas sensor array substrate.The thermal characteristics of different heating electrodes?Au,Cu and Pt?and structures?bottom etching,circumferential etching and hanging window structure?of the sensor unit are simulated and analyzed by finite element method.It was found that the average temperature of the sensing layer of the Pt heating electrode with the lowest thermal conductivity is the highest?336.01??,and the suspended window structure can greatly increase the average temperature of the center and the temperature difference between the center and the edge,it shows good thermal characteristics with low power consumption?16.71 mW,350??and fast response??100 ms?.Sensitive materials for gas sensor arrays were successed prepared by hydrothermal method.The characterization results show that rutile SnO₂-based metal-doped materials with morphologies of nanoflowers.Gas sensor array substrate?27 mm x 27 mm?was fabricated by silver paste fixation,wire bonding and droplet guiding coating methods.?3?Detaction characteristics of gas sensor array to individual gas in transformer oil were tested,and the results of first-principles calculation are verified and perfected.The temperature,concentration and response-recovery characteristics of metal-doped SnO₂-based gas sensor arrays for H₂,CO and C₂H₂ are obtained.It can be concluded that metal doping can reduce the working temperature of SnO₂,increase the concentration characteristics of gas?linear detection range and limit of detection?and accelerate the response speed.Among them,SnO₂ doped with Ag,Pd and Zn can effectively detect H₂,CO and C₂H₂ at concentrations below 0.5?L/L,respectively.By comparing the microscopic simulation results with the macroscopic detection characteristics,a metal doping modification mechanism was established,which was related to the band gap width and the optimum working temperature,gas adsorption energy and response-recovery characteristics,charge transfer and concentration characteristics.The accuracy and feasibility of the first-principles calculation for gas sensing mechanism analysis were verified.?4?Qualitative identification and quantitative estimation of multi-component gases by gas sensor array.Based on the gas sensor array,the mixed gas response tests of H₂,CO and C₂H₂ are carried out.A sample database of the gas sensor array response to the mixed gases is established.The response law of the sensor unit to the mixed gases is analyzed by the neighborhood algorithm?kNN?and data fitting method.It is found that the response of the sensor unit to the mixed gases presents a non-linear characteristic,and the fitting effect of the ternary power function is the best?R²>0.98?.The deep belief neural network?DBN-DNN?for qualitative identification and quantitative estimation analysis of multi-component gases is established.The results show that the DBN-DNN classification model can simplify the data preprocessing process with high classification accuracy?97.44%?.The average relative error of quantitative estimation of multi-component gases is only 5.37%.It can effectively suppress the cross-sensitivity of gas sensors in multi-component gases detection.The research results of this paper lay a foundation for on-line monitoring of dissolved multi-component gases in transformer oil based on gas sensor array.