Cloud Computing Platform Design and Machine Learning-Based Fault Location Method in Automatic Dispatching System of Smart Grid

In order to improve the effectiveness and efficiency of automatic dispatching of smart grid, fully exploiting the monitoring data and mining the inherent relation among the data are the key to the grid state monitoring, abnormality prediction and fast fault location for automatic dispatching service of the smart grid. With the rapid increase of grid scale and the type and volume of the monitoring data, distributed storage and computing-based cloud computing platform becomes the basic infrastructure of smart grid. In this paper, after the structure and function analysis of the current management and dispatching platform D5000, a cloud computing platform is designed and integrated into the D5000 platform. This cloud computing platform is constructed hierarchically, in which the Hadoop performs distributed data storage and computing via HDFS and MapReduce, while Spark implements data mining with the aid of Spark SQL when frequent data exchange and data computing is required. The data mining task includes modeling the state of the automatic dispatching subsystem, making early warning, and locating faults, for which machine learning-based algorithms are developed. The feasibility of the designed platform and the effectiveness of the proposed methods are verified.