【作者】 尹星光；

【导师】 莫耀赐；

【作者基本信息】 广西大学 ， 电力系统及其自动化， 2002， 硕士

【摘要】 本论文研究了人工神经网络在自适应输电线路距离保护中的应用。对基于人工神经网络的输电线路自适应距离保护进行了仿真研究和实际应用探讨。 距离保护受网络结构和运行方式的影响较小，因此距离保护长期以来一直是复杂电网中高压输电线路的主要保护方式，也是高压输电线路最重要、应用最广泛的保护方案。但是距离保护的性能也存在一些问题，当应用于超高压线路时，这些问题更为突出。为了解决传统距离保护的缺点，很多研究者都做了不懈的努力。本论文中，在前人研究的基础上，针对传统距离保护存在的反应经过渡电阻短路的能力这一问题，引进了神经网络，对这种新型的神经网络式的距离保护进行研究，利用神经网络的优越性，采用从系统中的各种参数的组合模式来识别出故障。该研究将综合反映上述各种特征，对电力系统的状态进行综合判断。通过大量样本训练后，将电力系统输电线路的各种运行状态信息记忆在其连接权值中，从而使该神经网络能正确识别电力系统输电线路的各种状态——正常运行、故障，自适应的调整距离保护安装处动作阻抗。 最后，利用PSCAD／EMTDC仿真广西百龙滩至恶滩220kV实际线路， 广西大学硕士学位论文·《基于人工神经网络自适应距离保护的研究》得到大量训练样本，再利用Mat ah押经网络工具箱对建立起来的神经网络对这些样本进行训练，训练结柬后，再在仿真系统里进行检验，仿真结果表明这种新型的输电线路距离保护装置能耿得令人满意的效果，很有研究和发展的价值，随着神经网络技术的发展，作为一种新型的距离保护装置，这种保护投入实际应用的前景也值得进一步探索和研究。更多还原

【Abstract】 In this paper, the application of artificial neural network (ANN) to distance protection of transmission lines in power system is presented in simulation and its practical possibility is also discussed.Although distance protection is the main protection in the protection of transmission lines in complex power network for it is not affected by the structure of the power network and the changing network conditions, its performance is not perfect and not appropriate to high voltage transmission lines. Some problems exist in the traditional protection method. In order to solve these problems, many researchers are working hard all the times. In this paper, ANN is introduced into this field to deal with the fault resistance problem. To resolve the capability problem in traditional device, this paper investigates how to identify the fault according to a different model of combining the various parameters in system. The estimation of the state of power system is finished after analyzing the combine model in terms of ANN. This research reflects all characteristics mentioned above andcomprehensively evaluates the state of power system. When the signals reflecting the state of power system is stored in the weights of ANN after training a great deal of samples, the output of ANN can exactly identify each state of the power system; and it can also adjust the starting impedance of the relay adaptively at the same time.Finally, this paper reports the simulative result of this new model of ANN. Numerous samples firstly acquired from simulating an actual transmission line -line Bailongtan to Etan of Guangxi power network with PSCAD/MTDC, and trained the ANN model established by the Matlab neural network toolboxes. According to these samples, the reliability of ANN model is verified. The satisfactory result of simulation shows that the adaptive distance protection based on ANN is promising and the research developing value of application will also make greater progress in the future. With the development of ANN technology, the scheme will be brought to center stage and make greater progress in the coming years.更多还原