Sensitivity Characteristics Of Oxide Chromatographic Detector And Its Application In Analysis Of Dissolved Gases In Oil

Power transformer is the key equipment in power grid.Monitoring the status of power transformers in real time by online monitoring system is of great significant to realize the safety and reliable running of power grid.Based on that,the most suitable plan for maintance can be determined,and the economic benefit of overhaul can be greatly improved.The chemical bond of insulating oil would be broken after transformer faults,hydrocarbons and other gases would be generated.The dissolved gas analysis?DGA?in oil plays an important role in guaranteeing the safe operation of transformers.In order to promote the development of condition-based-maintenance and online monitoring of oil-filled electrical equipment,e.g.,power transformers,in this paper,the tin oxide gas sensor,solid oxide fuel cell?SOFC?gas detectors and their gas chromatograph analysis system for DGA application are developed after theoretical simulation.The denosing algorithms of chromatographic signals are studied,as well as characteristics of the detectors,quantitative theory,fault diagnosis and state evaluation of transformers based on detected DGA data.Based on which,the online monitor for power transformer are implemented.The content and innovation of this paper are as follows:?1?By using first-principle method based on density functional theory?DFT?and taking H₂ as the representative gas of the dissolved gas in oil,the gas adsorption model with H₂ adsorbed on different characteristic atoms of Sn O₂?110?surface is established,and the adsorption energy is calculated theoretically before and after adsorption,as well as the Mulliken charge analysis,density states of H₂ and the partial density of the characteristic atoms.The theoretical calculation results show that the conduction band of Sn O₂ has a shift towards the Fermi energy after H₂ adsorption,which narrows the bandgap between the valence band and the conduction band,and increases the probability of the electron transition.O_2c is the best adsorption position of H₂.In the process of adsorption,charges transfer between the H₂ molecule and the surface.The electrons captured by the Sn O₂ surface are obtained by the Sn O₂,which increases the carrier concentration in the Sn O₂ conduct band,reduces the barrier height of the depletion layer of the Sn O₂ surface,and leads to the gas-sensitive characteristic.The above conclusion shows that the barrier theory can be used as the basic model of tin oxide in the quantitative analysis of gas in oil.?2?The tin oxide detector combined with gas chromatography technology is developed.The theoretical model of multi-component gas quantitative analysis based on tin oxide chromatography detector is proposed,which solves the problem of cross sensitivity and poor accuracy of linear calibration.Based on the electrospinning method,nano tin oxide materials and its corresponding gas sensors are developed.In order to detect the components of the gas mixture quantitatively and solve the cross-sensitivity problem,based on the simulation results and barrier theory of tin oxide semiconductor crystal,the oxygen ionization and adsorption-desorption model,the theoretical model of tin oxide chromatographic detector for quantitative analysis of multi-component gas is derived.Based on the genetic threshold wavelet signal denoising algorithm,the proposed theoretical model is validated.The experimental results show that the gas chromatographic analysis based on the tin oxide gas detector and its corresponding model proposed in this paper can accurately quantify the six-component fault gases?H₂,CO,CH₄,C₂H₄,C₂H₆ and C₂H₂?in oil without cross-sensitivity,which solves the problem of poor selectivity and accuracy of linear fitting when the tin oxide sensor is applied directly.The whole system uses cheap air as the carrier gas,the cost of the sensor as well as the chromatography system is low,which has important application value.?3?SOFC zirconia gas detector and its gas chromatographic analysis system are developed.Based on the analysis of SOFC gas sensing mechanism and Nernst theory,an N-S quantitative mathematic model and double logarithm fitting model of SOFC detector are proposed.The problem of large quantitative error of linear model with peak area has been solved,realizing the accurate quantification of dissolved gas in oil for different gas concentration ranges.In order to solve the problem of long stability time of tin oxide chromatographic detector and achieve high sensitivity analysis of trace gases dissolved in oil,the SOFC zirconia chromatographic detector and its corresponding gas chromatography analysis system are developed.The problem of large quantitative error with peak area linear fitting exists when the partial pressure of oxygen in SOFC detector changes.Based on the Nernst theory of SOFC as well as the assumption of full combustion of fuel gas,an N-S quantitative theoretical model based on oxygen consumption is proposed.When the concentration of combustible gas is relatively high,the double logarithm model?DLM?suitable for larger concentration gases quantitative analysis is proposed.The experiment verifies the proposed theoretical models in different concentration ranges.The results show that the developed SOFC chromatographic detector and the models proposed in this paper can accurately quantify the six-component fault gases in oil and are suitable for direct quantitative measurement of trace gas?<50?L/L?and relative larger gas concentration?>50?L/L?with DLM fitting method.The detection limit of acetylene can reach 0.05?L/L,which is conducive to the discovery of the early failure of the transformer.Quantative accuracy of the traditional peak area method is promoted largely when perform measurement of trace gases with N-S model directly.The entire system needs no calibration process,which improves the reliability and complexity of the hardware system.?4?A transformer fault diagnosis algorithm named RVM-ANFIS is proposed to solve the problem of low accuracy with the traditional intelligent diagnosis algorithm when diagnose the data with fuzzy characteristics.The purpose of the concentration measurement of dissolved gas in oil is to use the test data for transformer fault diagnosis and status evaluation.According the characteristics of DGA data samples,which are with finite sample,nonlinearity and fuzzines,a hybrid RVM-ANFIS fault diagnosis algorithm and its corresponding topological structure are presented to solve the problem of low accuracy and easy misjudgment of the traditional intelligent diagnosis algorithm when the characteristic of data fault is fuzzy.The procedure is:overheats and discharge faults are divided firstly with RVM,then ANFIS is used to classify the fault types further.Comparison of the conventional diagnostic algorithms,like ANN,SVM is performed to verify its effectiveness and superiority,and effectiveness of the algorithm's engineering application is introduced.The simulation results show that the misclassification of overheat and discharge fault is less than 1%when RVM algorithm is used.The RVM-ANFIS algorithm has strong adaptive ability,which is suitable for sample data with high dimension and small samples;the fuzzy method and membership degree method are especially suitable for the fault feature located in overlapping region.It provides an effective way for the fault classification and realization of the state evaluation for transformers,with an accuracy of 95%.