Study On The Detection Of Multi-components Gas Based On Nano-modified Graphene Sensor Array

Oil-immersed power transformer is one of the pivotal equipment in power system.Its safe and reliable operation directly affects the stability of power system.When oil-immersed power transformer is in a long-term stable operation state,its insulation system may suffer oil overheating,oil-paper overheating,oil-paper insulation partial discharge,oil spark discharge and other faults accompanied by faulty characteristic gases such as H₂,CO,CO₂,C₂H₂,CH₄,etc.Power transformer dissolved gas analysis?DGA?is one of the convenient and effective means for transformer early fault diagnosis.Gas sensor array detection technology is the core of analysis technology.Its detection accuracy and sensitivity will directly affect the correctness,reliability and life evaluation of transformer operation state analysis.Therefore,gas sensor array and its detection characteristics will be continuously studied and it is of great significance to improve the safe operation level of power transformer.In the key technology of gas sensor array detection,the classification and content detection of multi-component mixed gases is the key development direction.Among the reference fault gases for judging the operation status of power transformers,acetylene?C₂H₂?,hydrogen?H₂?and carbon monoxide?CO?are the main fault characteristic gases.Their contents can partially reflect the discharge and oil-paper overheating in oil-paper insulatiofaluts.Metal oxides?MOS?,such as tungsten trioxide?WO₃?semiconductor gas sensor,as one of the main gas sensors for laboratory research and commercial use have many advantages,such as small size,low cost and easy preparation.They are often used to detect mixed fault characteristic gases.However,their sensitivity is low and response is restored.The problems of long complex time,high working temperature and serious cross-sensitivity always restrict its further development.Since its discovery,graphene materials have attracted much attention because of their excellent electrical properties,huge specific surface area and high electronic mobility.It also provides a new idea and approach for developing high-efficiency and high-precision gas detection device and system for power transformer fault characteristic gas detection.In this thesis,six kinds of gas-sensing materials,namely,reduced graphene oxide?rGO?nanosheet,WO₃ nanoparticles,1 wt%,4 wt%WO₃-rGO,Pt-WO₃-rGO and Pd-WO₃-rGO,were prepared for three kinds of fault characteristic gases of power transformers.The gas-sensing materials were combined with gas sensor arrays optimized by self-assembly and spraying technology to respond to temperature and gas-sensing of gas-sensing materials.The selectivity and repeatability were tested.Based on the theory of electron depletion layer,the gas sensing mechanism of metal and metal oxide modified graphene gas sensing materials was explored,and the reasons for the excellent gas sensing response of the prepared materials were discussed in combination with the micro-morphology of gas sensing materials.Finally,the gas components to be measured were realized by combining the depth confidence neural network?DBN?model.And high-precision prediction of concentration characteristics.The main contents of this paper are as follows:?1?Aiming at the fault characteristic gases of acetylene?C₂H₂?,carbon monoxide?CO?and hydrogen?H₂?power transformers,six kinds of heterojunction-containing composite gas-sensing materials,namely pure rGO,pure WO₃ and rGO,were prepared by improved Hummer method,chemical reduction method,hydrothermal method,self-assembly method and thermal reduction method,respectively,with the mass percentages of 1wt%and 4wt%,Pd-WO₃-rGO and Pt-WO₃-rGO.The good morphology and successful doping of the prepared materials were proved by means of characterization.?2?Through the thermal field simulation module of COMSOL multi-physical field simulation software,the structure optimization design of interpolated micro-hot plate was carried out,and the optimal structure size was determined to be 1.5*1.5*0.3 mm,the heating area was 0.5*0.5 mm,the suspension film size was 0.6*0.6 mm,and the interpolation spacing was 25 um.The structure of interpolated micro-hot plate was obtained by combining micro-electro-mechanical technology?MEMS?and micro-electronics processing technology.Hybrid gas sensor array system with six micro-hot-plate units.?3?Six kinds of heterojunction-containing composite gas-sensing materials with pure rGO,pure WO₃ and rGO as WO₃ mass percentages were prepared by spraying method and self-assembly method,respectively,1wt%and 4wt%,Pd-WO₃-rGO,Pt-WO₃-rGO,respectively.The six kinds of gas-sensing materials were coated on the six-structure unit hybrid gas sensing array.Based on the laboratory micro-gas testing platform,the effects of six kinds of gas-sensing materials on acetylene?C-WO₃-rGO?were tested.Testing results show that pure rGO has no obvious gas-sensitive response to the three target gases.Pure WO₃ has better gas-sensitive response to the three target gases than pure rGO at 300 C.The gas-sensitive materials doped with 1 wt%rGO and 4wt%WO₃ have better gas-sensitive response to acetylene?C₂H₂?,respectively.Compared with pure WO₃,1wt%is the optimum doping ratio.1wt%WO₃-rGO has lower optimum operating temperature?150°C?,higher sensitivity and better gas sensitivity in detecting acetylene?C₂H₂?.At the same time,Pd-WO₃-rGO and Pt-WO₃-rGO prepared on the basis of 1wt%WO₃-rGO have better sensitivity and specific gas sensitivity than pure rGO,pure WO₃,1wt%WO₃-rGO and 4%wt WO₃-rGO respectively.That is to say,some gas sensitive materials prepared have better gas sensitivity to acetylene?C₂H₂?,carbon monoxide?CO?and hydrogen?H₂?.??4?The enhancement of gas-sensing response of graphene composite gas-sensing materials can be attributed to the p-n heterojunction structure formed between metal-doped semiconductor materials and rGO,which enlarges the specific surface area of the materials and effectively generates a large number of lattice voids,based on the microscopic characterization and gas-sensing reaction phenomena of the prepared 1wt%WO₃-rGO,4wt%WO₃-rGO,Pd-WO₃-rGO and PT-WO₃-rGO.Thus,it is more conducive to the adsorption and desorption of gases,and provides more active sites for gas sensing reactions.At the same time,the introduction of p-n heterojunction changes the conductivity of the material,making the material have higher carrier transport efficiency.?5?Aiming at the cross-sensitive response problem of six-unit hybrid gas sensor array system,based on the pre-processed experimental data sample base,the deep confidence neural network model is used to effectively reduce the cross-sensitive effect between six-unit hybrid gas sensor array system,and achieve high accuracy prediction of the components and concentration of the mixed gas to be measured.At the same time,it is found that DBN network is used to treat the mixed gas detection.The prediction accuracy of combined gas is higher than that of double hidden layer BP neural network model,which shows the validity and practicability of DBN neural network in transformer fault characteristic gas recognition and prediction.