论文部分内容阅读
电力系统高压同杆双回输电线的应用日益增多,但其故障识别与测距的问题尚未完全解决,同杆双回线因存在回路间耦合等因素,使得用单一的神经网络进行故障识别与测距的结果并不理想。作者比较分析了BP网络与Kohonen网络在同杆双回线测距方面的优缺点,提出了将故障识别与测距任务分配到多个网络的方法即将同杆双回线的每种故障模式各与一个BP人工神经网络对应,在线路上取一些固定点作为标志点,训练成功的BP网络输出的模糊值代表了标志点上发生故障的可能性。用模糊值构成插值曲线,根据曲线的相对位置确定故障模式,并由曲线的最小值求得故障距离。大量仿真表明该法可以准确可靠地确定故障模式并能测得较高的测距精度。
The application of double-circuit high-voltage transmission line with high voltage in power system is increasing day by day. However, the problem of fault identification and ranging has not yet been fully solved. Because of the coupling between two loops on the same tower, a single neural network Ranging results are not satisfactory. The author compares and analyzes the advantages and disadvantages of BP network and Kohonen network in the double-line distance measurement of the same tower. The method of assigning fault identification and ranging tasks to multiple networks is proposed. Each failure mode Corresponding to a BP artificial neural network, some fixed points are taken as marking points on the line. The fuzzy values output by the trained BP neural network represent the possibility of failure on the marking points. The fuzzy value is used to construct the interpolation curve, the fault mode is determined according to the relative position of the curve, and the fault distance is obtained from the minimum value of the curve. A large number of simulations show that the method can accurately and reliably determine the fault mode and can measure a higher accuracy of ranging.