Abstract:Transmission line is the key part of power system, the application of on-line monitoring technology of transmission line produces lots of data, and the deep excavation of the data becomes the hot spot of the big data research of power system at present. With the further application of smart grid data, a new solution is provided to ensure the reliable of power system. On the basis of studying the data type, data characteristic and data requirement of the on-line monitoring of transmission line, the Hadoop monitoring data model design that conforms to the characteristics of the large data structure of smart power grids is proposed. It contains three levels of multi-dimensional data input, distributed data storage, and distributed data processing.By building a big data processing environment based on Hadoop clusters, the PCA-SVM clustering algorithm is implemented under MapReduse parallel computing mode, and the fault type identification of transmission lines is taken as an example, transmission line fault identification based on data analysis is realized, which verifies the feasibility of this model to realize the online monitoring of transmission line.