A smart on-line over-voltage layered identification system

This paper proposes an effective smart on-line over-voltage identification system based on a new layered recognition structure. The over-voltage is a main and great threat for the safety and stability of electric system. It´s very helpful to analyze and identify over-voltage for improving power system insulation coordination. Through analysis of filed over-voltage data recorded by over-voltage on-line monitoring system, an over-voltage layered identification structure is presented for eleven kinds of common over-voltage. The wavelet transform and singular value decomposition theory are used to extract features of over-voltage. In consideration of the requirements of over-voltage recognition, the least square support vector machine is employed as the classifier to build a layered identification model, and the grid search algorithm is adopted to optimize the parameters of support vector machine to improve classification rate. The filed data testing results show that this system is feasible and promising for engineering application.