研究目的
To address the problem of on-line dissolved gas analysis (DGA) of a power transformer using a Fourier transform infrared (FT-IR) spectrometer to develop an analysis instrument for detecting internal faults.
研究成果
The FT-IR spectrometer-based DGA analyzer developed in this study meets the detection requirements for oil-dissolved gas analysis, with detection limits below 0.1 μL/L for all analytes. The novel approach of using two cone-type gas cells improves dynamics and reduces the effects of gas in gaps, enhancing stability and accuracy. The instrument's maintenance-free nature and high stability make it an ideal solution for on-line transformer fault detection.
研究不足
The study acknowledges the potential for baseline drift and abnormal distortion in the IR spectrum, which could affect long-term performance. Additionally, the presence of interferents like C3 gases and water vapor may impact analysis accuracy. The method's effectiveness in detecting H2 is limited, requiring a special sensor for accurate measurement.
1：Experimental Design and Method Selection:
The study utilized an FT-IR spectrometer to analyze dissolved gases in transformer oil, focusing on CO, CO2, CH4, C2H6, C2H4, and C2H2 as analytes, with C3H8, C3H6, C3H4, n-C4H10, and iso-C4H10 as interferents. A novel approach involving switching two cone-type gas cells into separate light-paths was introduced to reduce the effects of gas in the gaps between gas cells and spectrometers.
2：Sample Selection and Data Sources:
The instrument was tested with a standard gas mixture extracted from insulation oil in a power transformer.
3：List of Experimental Equipment and Materials:
The setup included an FT-IR spectrometer (Spectrum Two by Perkin Elmer), cone-type gas cells, temperature and pressure sensors (BMP085 by Bosch), and a vacuum degasifier.
4：Experimental Procedures and Operational Workflow:
The process involved scanning absorption spectra with two gas cells, calculating absorbance spectra, and analyzing the results with developed software to determine gas concentrations.
5：Data Analysis Methods:
The analysis employed baseline corrections, feature extraction methods (Tikhonov regularization and forward selection), and multi-input linear and polynomial models for component concentration determination.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
