Research On Grid Intelligent Alarming

The development of computer and ICT has greatly enhanced the level of grid monitoring and control, and it also has brought great convenience for dispatchers. However, with the expansion of the scale grid and the complex of connecting, a new question "How well understood so much information" comes out. Particularly, when emergency happens in power system, alarms swarms into the Control center, and the dispatcher can not understand the alarms in short time, so he also can't take right step to control. After several major power outage through out the world, engineers realized the need to use a computer to replace the dispatcher's work. So far, there are several alarm processors have been developed and used online over the world, however, due to the difficulties of maintenance and low work efficient, the existed alarm processors can not meet the whole world needs.This research focus on the drawback of traditional grid alarm processors, the mian content as followings:A new method which avoids the analog alarms processing is proposed in this thesis . The switch alarms are taken as the trigger source of the events and the events are distinguished according to the status of the node's voltage.The SVM classifier is used to make the system have the ability of "learning". Because it detects the relations between grid voltage vectors and events, without artificially rules, it is also good at reasoning. Besides, due to it avoid the inherent shortcomings of analog alarms, the processing speed also has been great speed up when compared to the traditional method.Through the simulation, we found that the ordinary SVM grid alarming method is inefficient when the events increase; So, we improve the method from two points. Considering the number of event types and the size of the grid have great impact on the processing speed, a novel alternative technology called FGIA(Fast Grid Intelligent Alarming) was proposed. PMU optimal disposition algorithm was used to reduce the dimensions of the observer, what's more, DT-SVM was also used to reduce thenumber of combination. A well improvement has been carried out after the development of the method.Finally, a grid intelligent alarming system based on FGIA method was designed. Modular design concept, as well as multi-threaded programming framework was used in this system. A part of functions have been achieved by using JAVA language.