Online Condition Monitoring of Overhead Insulators using Pattern Recognition Algorithm

In this paper commonly-used overhead line insulators are experimentally studied to propose an intelligent diagnosis system (IDS) to correctly identify the insulator health condition in real-time. The proposed IDS is developed based on three diagnostic indicators, third to fifth harmonic ratio of the insulator's leakage current (LC), cosine of the phase angle of the LC fundamental component, and the ratio of the maximum electric field stress to flashover electric field of the insulator. The proposed diagnostic approach can identify the normal, abnormal and critical conditions of an insulator based on the above-mentioned indicators. Leakage current and flashover voltage are experimentally measured for the studied insulators under various health conditions. Then, recorded data are analyzed to calculate the proposed indicators corresponding to each insulator state. Measured and calculated data are used to intelligently quantify threshold limits of each indicator based on visualization algorithm. In this algorithm, different classifiers are trained with experimental data. The decision tree, which provided the highest precision, is employed to determine reference boundaries of the three indicators for each health state of the insulator. Embedded sensors can measure and sample the LC and electric field stress of the insulator. Sampled data are transmitted to a central processing unit-based receiver via a radio communication channel to automate the identification of the insulator state in real-time.