Association Rules Mining Of Power System Transient Stability Assessment

Data mining technology is a promising and flourishing frontier in artificial intelligence field recently, which can extract useful information from huge amounts data set that is incomplete and noisy. Compared with classical theory of power system transient stability assessment, data mining involves more advantages whether in prevision or calculation efficiency. First, "Description to data overall characteristic and the forecast to whose developing trend" exactly is one of the data mining advantages. Second, data mining only needs some input parameters, not the entire description of the model, which might improve the efficiency and real time decision-making. As one of the most important methods of data mining, association rules technology could find the lurking rules of different attributes. Therefore, association rules technology is applied in power system transient stability assessment in this thesis.According to IEEE16 machine system, transient stability simulating is carried out. Meeting the data and the association rules algorithm demand, the paper provides a data mining preprocess study. It adopts FCM to adjust the number of the discretization points and use information entropy technology making continuous attribute value discrete. All the processes are illustrated by matlab programming. Moreover, the usual association rules algorithms, such as the classical Apriori algorithm, involve only one dimension. And hence, based on the deep research and study of Apriori, this thesis puts forward an improved algorithm well suited for the mining of multidimensional association rules. Finally, the thesis finds the association rules of different attributes and then might provide useful and helpful knowledge for power system transient stability assessment and predication.