具有冗余神经元神经网络模型系统的输电线路故障测距的研究

STUDY OF FAULT LOCATION FOR HIGH VOLTAGE OVER-HEAD TRANSMISSION LINE USING NEURAL NETWORKS MODEL SYSTEM WITH REDUNDANT NEURON

提出了一个层次化与模块化相结合的具有冗余神经元的神经网络 (NN)模型系统 ,该系统充分利用了神经网络在模式识别、非线性拟合及联想记忆等方面的优势 ,其模块化结构与生物神经网络功能区域结构相一致 ,信息处理机制符合生物神经网络分类和逐步推理的规律。该系统可实现高基金项目 :国家自然科学基金资助项目 ( 598770 1 6)。压 (超高压 )架空输电线路故障测距所需的复杂信息处理要求 ,可避免常规测距方法中出现伪根、迭代不收敛、及消除对端系统运行方式和助增电流影响导致测距误差大等不足。大量的电势暂态程度 (EMTP)仿真测试表明 :该方法的故障测距精度高、综合性能好、适应性强。

This paper proposes a Neural Networks(NN) system with redundant neurons based on the integrated module architecture and hierarchy architecture.This NN system adequately uses the powerful function of artificial neural networks at aspects of pattern recognition, nonlinear approaching, associative memory et. Its module architecture is similar with the function areas in human biologic NN system. And its information processing mechanism is consonant with the processing law of classification and step by step reasoning.This system can not only deal with the complex information required by fault location for HV overhead transmission lines, but also accurately locate the fault sites. So this method for fault location presented in this paper can eliminate the disadvantages in other conventional fault location methods, such as convergence to the false root or divergence in the procedure of iteration.And yet this method can eliminate the influences of load current and operating mode of the other terminal system, which result in the great location error in practice.Results from theory analysis and simulation by Electro magnetic Transient Program (EMTP) show that fault location precision of this method can completely satisfy practical requirements.