Reserch Of Separation And Detection Technology About Dissolved Gases In Transformer Oil And Development Of Experimental Platform

As the main power equipment, the reliable operation of the transformers is related to the security of grid safe operation directly in power system. Dissolved gas composition and content can reflect the internal transformer insulation fault effectively, so as one of the most popular and the most important technique in transformer online monitoring, reserching seperation and detection technology about dissolved gas in transformer oil is of great significance to improving the transformer fault diagnosis accuracy and reliability.In this paper, the technology current situation about the dissolved gas in oil online monitoring and the relation of transformer common faults and dissolved gas in oil is introduced firstly. Developed an experimental platform after making the analysis of the characteristics and composition of transformer online monitoring system, studied the transformer online monitoring from aspects of oil-gas seperation, experimental platform, mixed gas identification and experiments.Several common oil-gas separation methods are compared, and to adopt the polymer membrance method for oil-gas seperation in determined. The dissolved-diffusion process of gas in membrane is illustrated, combining the actual working condition of the transformer fault diagnosis, given the performance of the membrane being used. Introduced several common membrance characteristic. Conducted an experimental to verify several membrances permeability.On the basis of studying oil-gas separation technology, consider changing various factors, constructed oil-gas separation experimental platform. It is composed of 3 major subsystems: hydraulic circulatory system, oil temperature control system and oil-gas separation system. Given the basis of key components definition of each subsystem, the entire experimental platform construction provided a technical basis for studying oil-gas separation.Because of transformer fault gases are composed of mixed gases, related to information fusion technology, therefore, using gas sensor array combined with neural networks to analysis multi-gas composition. Introduced the gas sensor array and neural networks theory respectively, the reason that H₂ and CO can be used to reflect early internal insulation faults of transformer is given. Established the corresponding gas sensor array.Completed the calibration experiment of mixed gases identification, construction of neural networks structure. Trained and forecast the network using sample data from calibration experiment, analysed the error of forecast sample, the performance can meet the measure requirement. Designed the experimental procedure of oil-gas separation, effection of the membrance and chamber parameters on the oil-gas separation process is verified by experiment. The experimental results and theoretical analysis are comparable.