Research On Transformer Fault Identification Method Based On Multi-component Content Detection Of Free Gas

The power transformer is the core equipment in the whole power system,and it is also the key equipment for power transmission and voltage conversion.At present,the power transformer is still dominated by oil-immersed insulation.Mineral oil and insulating oil-paper will crack and produce H₂,CH₄,C₂H₂,C₂H₄,C₂H₆,CO,CO₂ and other gases dissolved in the oil under the discharge fault.However,in the case of sudden severe insulation failure,the gas can’t be completely dissolved in the oil,and some of the gas will diffuse in the form of free gas inside the transformer,resulting in a sharp increase in internal pressure.When the fault is serious,it will also lead to an explosion of the transformer.Although a series of studies have been carried out on dissolved gas in oil at this stage,the analysis method of dissolved gas in oil takes a long time and cannot reflect the fault state in a short time,so it is difficult to effectively diagnose the sudden discharge fault state.Therefore,in order to find out the sudden fault of the transformer,it is of great significance to construct the law between the multi-component content of free gas and the fault type.Aiming at the main problems of free gas generation inside the transformer,the process of gas production from mineral oil cracking,the content characteristics of free gas components and the fault identification method are studied by means of simulation and experiment.The main contents of this thesis are:（1）In this thesis,a model containing 20 mineral oil molecules was designed by molecular dynamics simulation software.Based on the Reax FF force field,the mineral oil model was simulated to explore the cracking reaction process of mineral oil at different temperatures under the same electrical stress.The number and types of small molecules generated at four different simulation temperatures were extracted by self-compiled program,and the rules between small molecules such as H₂,CH₄,C₂H₂,C₂H₄ and C₂H₆ and simulation temperature were explored,which provided a theoretical basis for studying the gas production law of mineral oil.（2）According to the typical discharge fault inside the transformer,a simplified device for simulating the oil tank of the transformer is developed,and a simulation test platform for transformer breakdown discharge gas production is built.Three different breakdown discharge models are designed.The voltage and current signals and gas volume were obtained in the experiment.The law between gas volume and breakdown discharge energy was explored,and the discharge energy under three different fault types was obtained.The component contents of dissolved gas and free gas in oil under three different fault types were detected by chromatographic analyzer.The component content characteristics of dissolved gas and free gas in oil were studied,which provided data support for constructing the law between multi-component content of free gas and different fault types.（3）The method of identifying different discharge fault types based on the multi-component content of free gas is studied.The particle swarm optimization algorithm is used to find the optimal solution of the penalty parameter C and the kernel function parameter g of the support vector machine,improve the recognition rate of the SVM,and construct the support vector machine classifier model optimized by the particle swarm optimization algorithm.The normalized spectra of seven kinds of data were drawn based on the contents of H₂,CO,CH₄,C₂H₄,C₂H₆,C₂H₂ and total hydrocarbons,and seven statistical features of the spectra were extracted for pattern recognition.The results show that the PSO-SVM classifier can effectively identify the statistical features of the normalized spectrum of free gas.