Research On Power Grid Intelligent Alarm System Based On Data Mining Technology

With the development of the power industry, the power grid is increasingly large and complex. A lot of information is reported to Grid Dispatching Center when there is anomaly or fault in the grid, and the information is shown to dispatchers directly and change quickly. The true and important alarm information can not be extracted and shown so that it can not really play the role of alarm because of lacking of logical reasoning. In addition, the abnormal and random errors of the remote system, the communications system and the related measuring equipment running result in alarm information distortion, which make the dispatcher can not determine the operating status of power grid currently. Therefore, intelligent process of alarm information and research of the power grid intelligent alarm system become the focus in the areas of network operation and control of domestic and international.The alarm information is mostly shown by classifying in the existing alarm systems, and denoising rate of the systems is not high. And alarm information is mostly displayed by light plate directly in the alarm system, which lack of logical reasoning and ignore the connection among the alarm information so that the dispatcher can not get the comprehensive warnings and the problems in the grid. In view of these problems, the Decision Tree method and the Rough Set method of the Date Mining technology are researched in this paper. On the basis of the research, the attributes of alarm information are analyzed to improve the model of decision tree for alarm information denoising, which increases the denoising rate for noise. And the power grid alarm rules are mining by Rough Set method to ensure completeness of the rules. The power grid intelligent alarm system is designed and researched based on above method in this paper. This system designed has been used in some dispatching automation system and has made remarkable effect.