基于神经网络的配电线路高阻抗故障检测方法的研究-电气工程专业论文.docxVIP

哈尔滨工业大学工程硕士学位论文 哈尔滨工业大学工程硕士学位论文 - - II - Abstract High Impedance Fault (HIF) of electric distribution systems is a great problem which can affect the safety and reliability of electric distribution systems. Not like short-circuit faults，a HIF usually draws insufficient current to be detected by conventional means, such as fuses and relays. So it is hard to detect a HIF. And if a HIF remains unsettled, it deteriorates the quality of electricity and is a waste of energy, even a threat to lives. This paper is financially supported by the Visiting Scholar Program of Chongqing University, purpose of which is to do research on the analysis and detection of HIF. A scheme is given to detect an arcing HIF of 10kV distribution systems in this paper. It adopts outdoor current transducer to get the phase current of distribution line, uses DSP to extract the characteristic vector and transmits the suspicious data through GPRS to a Server to judge whether a HIF has occurred. The current data sampling hardware and flow chart of DSP firmware is designed in this paper. The GPRS communication test is also done. Then the load model and HIF model of a single branch distribution line are designed to get training and testing data through computer simulation. After the characteristic vector is extracted, according to Pattern Recognition theory, a classifier based on Artificial Neural Network is established. This paper did simulations on the designed algorithm to detect a HIF. The results prove that the designed classifier can distinguish HIF from capacitor switching, load switching and saturable current transformers, correctly classifying a HIF. Keywords HIF; Pattern Recognition; Artificial Neural Network - - PAGE IV- 目录 摘要I Abstract II 第 1 章 绪论 1 1.1 课题来源及研究的目的和意义 1 1.2 高阻抗故障检测国内外研究现状 1 1.3 本课题主要研究内容 5 第 2 章 高阻抗故障检测系统设计 6 2.1 总体方案设计 6 2.2 数据采集终端硬件设计 8 2.2.1 电流传感器接口设计 9 2.2.2 数据采集终端控制器电路设计 11 2.2.3 GPRS 通信模块接口电路设计 12 2.2.4 电源系统设计 13 2.3 数据采集终端控制器软件设计 15 2.3.1 电流信号检测程序设计 16 2.