基于模糊隶属函数的变压器色谱峰定性算法

把模糊技术引入了电力变压器色谱峰定性领域。该算法为:把要定性的色谱峰位作为模糊矢量,把各个成分作为模糊集元素,峰位相对于成分的隶属度函数用成分的绝对保留时间来构造,不失一般性,隶属度函数的形状取为三角形,2个相邻成分绝对保留时间中点的隶属度值设为50%,若某个峰位相对于某个成分的隶属度函数值大于50%,则这个峰位就是所用隶属度函数的那个成分的峰位,若存在多个峰位隶属于某个成分,则隶属度最大的那个峰位是相应成分的峰位,其余是假峰。对该算法进行了测试,结果表明:该算法方法简单,抗峰漂移性能良好,可以剔除假峰,具有显著的优点。

Qualitative Algorithm for Fuzzy Membership Function Based Transformer Chromatogram Components

The fuzzy technology is introduced into the field of qualitative analysis of transformer chromatogram components. That is to say, with the chromatogram peak's position taken as a fuzzy vector, each of the components as a fuzzy set's element and the components' retention time as the membership function, every peak's position to every component is constructed. Without loss of generality, the membership function' s shape is taken as a triangle and the membership function value of the midpoint of the two neighboring components' retention time is taken as 50%. If the membership value of a certain peak's position with respect to its component is greater than 50%, the peak position's component is the used membership function's component; and if there exist many membership function values mapped onto one component, then the peak position with maximum membership is the corresponding component's true position, while the remainder are fake positions. It is shown by experiment that the method proposed is simple and the anti-peak-position-shift ability is evidently remarkable.