Skip to Main Content
In order to cope with extreme natural disasters which appear more and more frequently, it is necessary to extend the data acquisition range from traditional electric quantity of power systems to macroscopic and microcosmic meteorology data, such as those on thunderstorm, typhoon, ice and snow, as well as pollution flashover, etc. Due to the mechanisms' differences, the data needed for early-warning might be different for various types of natural disasters. Taking thunderstorm as an example, this paper analyzes the mechanisms of natural disasters affecting power systems. The required data are discussed and an early-warning method is proposed based on data provided by a lighting location system. Firstly according to the proposed criteria, the thunder locations detected within the current interval are grouped into areas; secondly the results are matched with those of the previous interval; finally the most possible thunder areas can be estimated for the next several intervals. Unlike traditional models with annual average probability of thunder-striking faults, the proposed real-time probability model which considers real-time information is space dependent, as well as time dependent. Instead of being fixed, the list of preassigned faults to be on-line assessed is time varying. Thus, the task of online security analysis is adaptive and more efficient.