Crucial Power Flow Interface Discrimination Based on Distributed Improved-SVM Classification in a Big Data Set

The operational states of power system become much more complicated and variable, since the size of power system grows larger. To secure power system operation, crucial power flow interfaces should be monitored. Therefore, a power system safe operation knowledge base should be established to help system operators to make decisions. The base works like an automatic operator (AO), which can quickly discriminate the crucial power flow interfaces according to power system real-time operation conditions. In this paper, first a description of the classification problem is given. Next, models of conventional support vector machines (SVM) and increment SVM are briefly described. Then the distributed computing framework of the power system safe operation knowledge is designed and the knowledge base can be established and updated based on improved-SVM. Finally, the application of the knowledge base in Guangdong Province Power System in China shows its advantages in accuracy and classification speed.