Using support vector machines for determining voltage unstable areas in power systems

This paper presents the application of support vector machines (SVM) for determining voltage unstable areas in an actual power system. The voltage unstable area is first determined based on the power transfer stability index (PTSI) calculated using information obtained from dynamic simulation output. Simulations were carried out on a practical 87 bus test system by considering load increase as the contingency. The data collected from time domain simulations are then used as inputs to the SVM which acts as a classifier to determine the voltage unstable areas in the power system. To reduce training time and improve accuracy of the SVM, the Kernel function type and Kernel parameters are considered. To verify the effectiveness of the proposed SVM method, its performance is compared with the learning vector quantization (LVQ) technique. Studies show that the SVM gives similar classification accuracy as the LVQ with 100% accuracy. In terms of computational time, the SVM is faster than the LVQ.