作　　者： ; ; ; ; ;

机构地区： 武汉大学电气工程学院

出　　处： 《中国电机工程学报》 2007年第16期19-23,共5页

摘　　要： 应用信息融合技术对电力系统暂态稳定进行评估,提出了一种基于“结合人工神经网络和D-S证据理论的信息融合模型”的电力系统暂态稳定评估方法。当电力系统发生故障时,采用该方法可以综合来自电网和发电机的多个信息源对电力系统的暂态稳定进行判别。首先,选取能迅速反映电力系统暂态过程的特征量,如功角、加速度、转子动能等,进行像素级融合。然后,将这些特征量划分为时间和空间2个征兆域,并分别输入至设定的3个子神经网络进行特征级融合。最后,将特征级融合的输出作为决策级融合的输入,利用D-S证据理论实现时间域和空间域的决策级融合,从而提高了电力系统暂态稳定评估的可靠性。利用中国版BPA暂态稳定程序和电力系统全过程动态仿真软件,对7机24节点系统进行了仿真。仿真结果表明,基于信息融合技术的电力系统暂态稳定评估方法较基于人工神经网络的评估方法更为快速、准确。 A power system transient stability assessment method is proposed based on an information fusion model of combining artificial neural net （ANN） method and D-S evidence theory in the paper. When accidents occur, much information obtained from power networks and generators are synthesized to diagnose power system transient stability by the studied method. Firstly, the characteristic parameters that can rapidly reflect the transient process, such as power angle, acceleration and rotor kinetic energy, etc., are selected for image-level fusion of ANN. Then the characteristic parameters are classified into time and spatial symptom field, and inputted respectively to three predefined neural subnets for feature-level fusion. Finally, the feature-level fusion outputs are used as the inputs of decision-level fusion, and fused in time and spatial domains by means of D-S evidence theory for further reducing the uncertainty of transient stability assessment. A 7-generator and 24-bus power system was simulated by China version BPA （Bonneville power administration） Transient Stability Program and the Power System Full Dynamic Simulation Program. The simulation result indicates that the proposed method is more precise and rapid than the ANN method under the same conditions.

关 键 词： 特征量 信息融合 人工神经网络 证据理论 电力系统暂态稳定分析

领　　域： [电气工程]