Imbalanced data processing based on adaptive composite sampling algorithm in the application of High Voltage Circuit Breaker maintenance

摘要

It is important to make sure that the High Voltage Circuit Breaker (HVCB) have reliable performance in power system. So it is important to have a knowledge of when the HVCB need maintenance. The paper would use the data mining techniques to predict when to maintain HVCB. As the data is imbalanced, one class is less than the other, the paper would like to introduce an imbalanced learning processing algorithm, called ACSA. The core idea of ACSA to solve when to give a maintenance is, according to the difficulty of the study sample, sample for each of the different minority assign different weights. As a result the experimental analysis shows that the ACSA approach improves learning by improve data distribution.