Estimation of the Degree of Polymerization (DP) of Oil-Paper Insulated Transformers Using a Novel Fuzzy Logic Algorithm

Inhalt:
The most common parameter used for the determination of the condition of the widely used paper insulation is the Degree of Polymerization (DP). Since a paper sample is necessary for the DP measurement, it is not applicable for the determination of the paper quality of components that are in service. However, there are several approaches to derive the DP value without a direct measurement, e.g. analysing the fluid, which is used to impregnate the paper, thus typically an oil paper insulation is used in many HV components such as transformers. Today an estimation of the DP value is often done using the correlation between the furan content in the oil and the degradation of the paper. However, this of-ten leads to a not accurate result, especially in cases, in which e.g. a local overheating creates at one spot severe deterioration of paper and therefore a high furan content. In such cases the furan content does not correlate with the average DP value of the whole insulation system, because it is a result of strongly aged paper at a single location only. Therefore, additional indirect methods are needed in order to obtain more reliable results regarding the DP of the whole insulation system. In this contribution therefore, a novel approach for predicting the DP value of paper insulation based on physicochemical characteristics is presented. Parameters like Acidity and Interfacial Tensile (IFT) of the insulating oil as well as Dissolved Gas Analysis (DGA) results, like the concentration of carbon monoxide (CO) and carbon dioxide (CO2) are evaluated by a special Fuzzy Logic System. A novel algorithm is implemented to create the rules for the fuzzy sets based on the information gain according to the entropy of the measured data as a measure to determine the data set characteristics.