基于神经网络的自适应继电保护的研究

【作者】 邓烽；

【导师】 都洪基； 张俊芳；

【作者基本信息】 南京理工大学 ， 电力系统及其自动化， 2002， 硕士

【摘要】 随着电力系统的发展和对安全运行要求的提高，常规的继电保护和故障诊断技术已经不能适应不断发展的系统要求，因此自适应继电保护越来越受到重视，并已有广泛的研究。本文从分析传统继电保护的缺点入手，运用人工神经网络所具有的自适应能力、学习能力和模式识别能力，提出了一种实现自适应电流保护的神经网络模型。该保护不仅能够随着运行方式和故障类型的变化自适应的改变故障判别最佳条件，还能在发生任何故障时准确判别故障方向，没有死区存在。同时对系统振荡和过渡电阻对保护的影响进行了研究，提出了一种自适应距离保护模型。为了验证其可行性，对两个模型分别进行了大量的仿真计算分析，验证了所提方法的可行性与优越性。更多还原

【Abstract】 This paper describes a new method of the application of ANN in the adaptive relay protection .It means that some problems existing in the power system, i.e faults analysis, direction distinction could be solved by using strong adaptive ability and studying ability of ANN. This protection includes two independent subnets to finish their functions in parallel way and has an advantage that the relay setting values are determined according to current operation condition of power system and fault type. It can keep the protection characteristics optimal. In order to deal with oscillation and fault resistance, the paper presents a new method of adaptive distance relay protection based the above relay protection. We made simulated calculation on high voltage transmission system and trained the nets finally, the result show that this method is absolute reliable using in the protection of power system.更多还原