摘要
Before diagnosed by DGA (dissolved gas analysis) methods, gas caution values, which index the level of gas formation, must be used to evaluate the possibility of incipient faults to reduce the misdiagnosis in the normal state. However, the calculation of these values is now only based on cumulative percentile method without taking into account operating conditions. To overcome this disadvantage, a new approach to calculate the transformer caution values is presented. This approach is based on statistical distribution and correlation analysis, and it takes the individual variation and fluctuation caused by internal and external factors into consideration. Then 6550 transformer DGA data collected from North China Power Grid are analyzed in this paper. The results show that the volume fraction of TH (total hydrocarbon) approximately obeys normal distribution when the 3-sigma rule is used to calculate its caution value. The volume fraction of CO has a strong positive correlation with oil temperature. For H2, the negative correlation with oil temperature is significant when the volume fraction is not very low. The caution value curves for CO and H2 are obtained by regression analyses. Thus, the gas caution values/curves obtained using the new method are not always constant, but vary with oil temperature, which is an advantage of the proposed method compared with cumulative percentile method. The variation of gas caution values/curves also reflects the influence of the external factors, for instance, va- rying with monitoring time ensures that the gas caution values are always consistent with operating status.
Before diagnosed by DGA (dissolved gas analysis) methods, gas caution values, which index the level of gas formation, must be used to evaluate the possibility of incipient faults to reduce the misdiagnosis in the normal state. However, the calculation of these values is now only based on cumulative percentile method without taking into account operating conditions. To overcome this disadvantage, a new approach to calculate the transformer caution values is presented. This approach is based on statistical distribution and correlation analysis, and it takes the individual variation and fluctuation caused by internal and external factors into consideration. Then 6550 transformer DGA data collected from North China Power Grid are analyzed in this paper. The results show that the volume fraction of TH (total hydrocarbon) approximately obeys normal distribution when the 3-sigma rule is used to calculate its caution value. The volume fraction of CO has a strong positive correlation with oil temperature. For H2, the negative correlation with oil temperature is significant when the volume fraction is not very low. The caution value curves for CO and H2 are obtained by regression analyses. Thus, the gas caution values/curves obtained using the new method are not always constant, but vary with oil temperature, which is an advantage of the proposed method compared with cumulative percentile method. The variation of gas caution values/curves also reflects the influence of the external factors, for instance, va- rying with monitoring time ensures that the gas caution values are always consistent with operating status.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2013年第8期1960-1965,共6页
High Voltage Engineering
基金
Project supported by National Basic Research Program of China (973 Program) (2009CB724508)
关键词
溶解气体分析
统计分布
变压器
相关性分析
计算
体积分数
运行状态
操作条件
transformer insulation
caution values calculation
DGA
regression analysis
distribution regularity of volume fraction
oiltemperature