330 kV变电站过电压仿真分析及分类识别

Simulation Analysis and Classification Recognition of 330 kV Substation Overvoltage

随着电网的不断扩展,电力系统过电压对输电线路和电气设备造成的危害日益严重,研究电网过电压信号对有效遏制过电压和完善系统绝缘具有关键意义。以某330 k V变电站为例,基于ATP-EMTP构建过电压仿真模型,取得单相接地、空载变压器合闸、电容器组合闸、空载线路合闸、直击雷以及感应雷6种典型过电压波形;然后采用时域与小波分析法提取电压有效值、陡度、小波能量等14个具有良好差异性的特征量;构造了基于SVM的过电压种类识别结构,用以对过电压特征量在每层SVM中进行选择。仿真计算结果表明,上述6种过电压平均识别率为87.9%,识别速率约1 s。该方法识别精确率高,运行速度快,可满足对电网过电压自动识别的工程要求。

With continuous expansion of power grid scale, overvoltage of power system is causing serious and increasing damages to transmission lines and electrical equipment. There-fore, research on the power system overvoltage signals is of great significance for effectively restraining the overvoltage and improving the insulation coordination. Based on ATP-EMTP and taking a certain 330kV transformer substation as an example, this paper establishes an overvoltage simulation model from which 6 kinds of typical overvohage wave forms like the single-phase grounding, no-load transformer closing, capacitor banks closing, no-load line closing, direct lightning and induced lightning are gained. Furthermore, 14 features of striking differences such as voltage effective value, steepness, wavelet energy and other features have been analyzed by the time domain and the wavelet analysis method. At last, the paper constructs a system that can recognize different overvohage layers and make it possible to select from each overvoltage based on SVM. The results show that the average recognition system accuracy for the 6 kinds of overvohage can be up to over 87.9% and the recognition speed is about ls. The recognition system is not only accurate enough but also very fast, and it can be used to meet the demands of the auto-recognition of the overvohage in the grid engineering practice.