Development of Insulator Diagnosis Algorithm Using Least-Square Approximation

This paper presents a new insulator diagnosis algorithm that is applied to the measured voltage and resistance values of an insulator string to detect faulty insulators. Under normal conditions of low pollution and low humidity, the distribution voltages of a sound insulator string show a U-shaped distribution, and its insulation resistances show very high values of above 20 GΩ. The inspection data measured in the field, however, showed that in high pollution or high humidity environments, the voltage distribution has sawtooth shapes and offsets in comparison with those of other strings, and its insulation resistances are greatly decreased. For these reasons, the commonly used algorithms tend to yield incorrect results under these conditions. In order to solve this problem, we propose a new diagnosis algorithm that can precisely detect faulty insulators from the measured data, regardless of the environmental conditions. We confirmed its effectiveness through live-line field tests in actual power transmission lines.