Social Network Power System Alarm Application

Millions of people share their thoughts and experiences instantaneously through social media. This could be a good source of information for power system operators to glimpse what is going on with their equipment before something bad really happens to the system such as a wildfire near power lines. Gaining a few minutes of head start could save a tremendous amount of time and effort if it could help prevent a fault or even reduce the impact of a fault in the system. This paper proposes a real-time warning system using the information gathered from Twitter to detect a possible threat to the grid. The collected tweets are processed by a machine learning algorithm to determine the possibility of threats to the system in their message. The training data for machine learning consists of six thousand tweets. The confidence of the machine learning is calculated based on the distance of the potential threat, the frequency of similar tweets being posted in a short period of time, likes, retweets, and some other factors. The tweets with a high possibility of threat are displayed on a website to give a warning so the system operators can react to them as early as possible. The proposed system is evaluated with real-time fake tweets to show its validation.