TRANSFORMER PAPER CONDITION ASSESSMENT USING ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM (ANFIS)
Because of aging, the paper insulation loses its mechanical strength and its ability to withstand electrical faults decreases drastically. The life of transformer is the life of paper, because paper insulation can not be reconditioned like oil insulation. Furan is a parameter that indicates the cond...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Subjects: | |
| Online Access: | https://digilib.itb.ac.id/gdl/view/81336 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Because of aging, the paper insulation loses its mechanical strength and its ability to withstand electrical faults decreases drastically. The life of transformer is the life of paper, because paper insulation can not be reconditioned like oil insulation. Furan is a parameter that indicates the condition of paper insulation because it can directly be transformed to degree of polymerization estimated (DPest), and has a high stability in transformer oil. In Indonesia, the transformer furan data is limited, so it is necessary to find the alternatives for assessing the transformer insulation paper condition using other widely available test parameters.
CO and CO2 have long been recognized as one of the cellulose degradation products, while interfacial tension (IFT), acidity, and color from oil insulation are statistically showing correlation to DPest. In this research two models have been made, which is statistic model using Multiple Regression (MR) and artificial intelligence model using Adaptive Neuro Fuzzy Inference System (ANFIS). The prediction results show that the ANFIS model can be used to predict the aging rate of transformer paper insulation, and has a higher accuracy than using the MR model where MAE (Mean Absolute Error), SMAPE (Symmetric Mean Absolute Percentage Error), and RMSE (Root Mean Square Error) obtained from ANFIS model prediction is lower. Accuracy of ANFIS model proposed for predicting transformer paper condition is 85.75%.
The output of this research is a ANFIS model based software that can be used to predict the condition of transformer paper as a furan substitution using CO, CO2, interfacial tension, acidity, and color as input parameters. In addition, %Eprl (Estimated Percentage Remaining Life), end of life approximation, and overall transformer dielectric conditions are also being assessed. |
|---|