Abstract: In order to achieve accurate early-warning of network public opinion for electric power grid emergency, various kinds of influencing factors leading to the network public opinion of electric power grid emergency were gotten with the information security risk assessment theory, and a novel network public opinion early-warning index system was also proposed based on support vector machine. The designing process of early-warning system was abstracted into the classification problem by the way that the values of all indices were treated as the inputs of classification, and the early-warning levels were treated as the outputs of classification. Support vector machine was used to solve the classification problem, which can effectively avoid the subjective shortcomings of the traditional methods. Finally, the validation experiment based on the network monitoring data of the Shenzhen blackouts on April 10, 2012 shows that this support vector machine-based early-warning system has good performance in time consuming and alarm accuracy.

Key words: electric power grid emergency, network public opinion early-warning, network public opinion early-warning system, classification problem, support vector machine