Decision Tree combining with the Attribute Reduction Algorithm for Substation Alarm Processing

In order to process the uncertain information of the substation alarms, the traditional Decision Tree Method is improved. The redundant signals in alarm sequence are simplified with the algorithm based on Positive Domain Attribute Reduction, the result of which is used as the characteristic attribute of the Decision Tree. Meanwhile, by constructing the Decision Tree in cooperated with temporal logic relation of the alarm signal, the characteristic attribute sequence can be accurately sorted, which finally produces a reasoning result for the alarm analysis. The method proposed in the paper is of concise and reasonable, and is proved having a good effect for fault classification decision making of the substation alarm processing.