Contamination Level Prediction of Insulators Based on the Characteristics of Leakage Current

• Published in: IEEE transactions on power delivery Vol. 25; no. 1; pp. 417 - 424
• Main Authors: Li, Jingyan, Sun, Caixin, Sima, Wenxia, Yang, Qing, Hu, Jianlin
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.01.2010; Institute of Electrical and Electronics Engineers
• Subjects: Applied sciences; Contamination; Contamination flashover; contamination prediction; Density; Electrical engineering. Electrical power engineering; Electrical power engineering; Equations; Exact sciences and technology; Flashover; Humidity; Insulation; insulator; Insulator testing; Insulators; Leakage current; Mathematical models; Miscellaneous; neural network model; Neural networks; Optimization; Overhead networks; Pollution; Power networks and lines; Power system reliability; prediction characteristics; Relative humidity; Surface contamination; Various equipment and components; Relative humidity; Insulator; Electric energy transportation; Contamination flashover; prediction characteristics; Electric discharge; Learning; neural network model; Salt; Pollution; Electrical flashover; Alarm system; Standard deviation; Safety; Root mean square value; Fog; Regression equation; Dielectric materials; Surface line; Neural network; Mean value; contamination prediction; Leakage current; Reliability; Power transmission line
• ISSN: 0885-8977; 1937-4208
• DOI: 10.1109/TPWRD.2009.2035426

In order to assess how severe the contamination level of the surface of power line insulators and to prevent unpredictable contamination flashovers, it is important to seek optimal prediction characteristics. That leads to the increase of the warning time and to the improvement of the reliability of the pre-warning system. Nearly 30 insulator strings at five pollution levels were tested in an artificial fog chamber, where their leakage currents were continuously recorded at the same operation conditions. The three characteristics of the leakage current, namely the mean value, maximum value, and the standard deviation of the root-mean-square (RMS) value of the leakage current, have been extracted. They describe jointly the current contamination levels of an insulator surface. In addition, regression equations between the three characteristics and various contamination levels have been established. The same three characteristics have been selected and used as the inputs of a neural network model together with two more parameters, the relative humidity and operating voltage. Also, the influence of each characteristic on the contamination prediction results has been investigated. The model is appropriate to predict the equivalent salt deposit densities (ESDD) with a difference of less than 0.035 mg/cm 2 if the training data and the testing data are selected at the security stage. This research results in the optimal prediction input parameters and sufficient pre-warning time before a contamination flashover.