8th International Conference on Advanced Power System Automation and Protection (APAP2019)

837 / 2019-04-29 22:00:19

A Health Risk Prediction Method for Power System Equipment Based on the Stochastic Forest Decision

Health Risk Prediction; Power System Equipment; Stochastic Forest Decision; Power equipment security

With the proposal of ubiquitous power Internet of things, a large number of terminal devices as well as users will be access to the power grid in the future. With the gradual evolution of closed power grids into open and interactive power Internet of things, the security of power terminals is facing unprecedented challenges. In order to meet this challenge, this paper presents a novel method for security risk assessment of power system equipment. According to the temperature, power consumption, message flow, CPU load rate and other information collected by the equipments, random vectors are constructed through random selected information, then multiple decision trees are constructed based on the random vectors. Based on the majority voting mechanism of stochastic forest decision-making, the health risk of power system equipment is online analyzed and predicted, which can help to timely report the early warning information, so as to avoid serious failure of the power system equipment.