基于多信息源的多层次配电网故障诊断方法研究

发布时间：2019-01-11 09:36

【摘要】：随着我国现代化技术的不断发展和人们生活水平的不断提高,电力已经逐渐成为人类生活不可或缺的一部分。配电网是电力系统的重要环节,其安全性与可靠性是保证电力系统稳定运行的关键。电网一旦发生故障或遭受威胁,若不能及时地进行故障处理,将极大地影响用户的供电质量,甚至会带来巨大的经济损失,而处理的前提是故障诊断,目前的配电网故障诊断方法多数仅依据开关和保护动作的信息,然而在开关或保护出现拒动、误动或者复杂故障情况时,仅靠单一的数据源的传统方法已经不能够准确的诊断出故障元件。因此,研究充分利用多数据源的配电网故障诊断方法具有十分重要的意义。本文通过分析目前电力系统故障诊断方法和故障信息源的研究状况和存在的问题,提出了基于多信息源的多层次配电网故障诊断方法。首先从故障数据信息获取的难易程度及多少的角度出发,提出一种基于多源信息配电网故障诊断方法。包括用于快速诊断简单故障的开关层、着力解决开关和保护异动情况下复杂故障的馈线层、以及准确判断复杂系统环境下多类型故障的变电站层,同时采用动态跳转策略,依据故障特点纵向调整诊断入口和结构。本方法增强了各层诊断的适应性,提高了故障诊断的效率与精度。其次从实际配电网系统角度出发,提出一种基于模糊Petri网的分层多源信息融合的故障诊断方法。该方法包括初步确定故障候选集的开关层和采用基于方向性加权模糊Petri网、小波变换技术及改进D-S证据理论融合的多源信息融合诊断层。该方法能够自动适应网络拓扑变化,具有较好的通用性和容错性。最后考虑配电网故障类型的准确辨识是实现故障定位的前提,在综合分析了配电网故障特征之后,提出基于免疫神经网络的配电网故障类型识别的方法。该方法利用小波变换技术提取故障特征作为免疫神经网络的输入,再利用免疫算法的全局搜索性能对神经网络的权值进行优化,避免了神经网络算法陷入局部小点,该方法能够较准确地实现各种故障模式下得故障类型识别,具有较好的鲁棒性和收敛性。
[Abstract]:With the development of modern technology and the improvement of people's living standard, electric power has become an indispensable part of human life. Distribution network is an important part of power system, its security and reliability is the key to ensure the stable operation of power system. If the power network fails or is threatened, the quality of power supply will be greatly affected and even huge economic losses will be caused if the fault treatment is not carried out in time, and the premise of the treatment is fault diagnosis. Most of the current fault diagnosis methods of distribution network are based on the information of switch and protection operation, however, when the switch or protection occurs in the case of failure, maloperation or complex fault, Traditional methods based on a single data source can not accurately diagnose fault components. Therefore, it is very important to study the fault diagnosis method of distribution network using multiple data sources. Based on the analysis of the current research situation and problems of fault diagnosis methods and fault information sources in power system, a multi-level fault diagnosis method for distribution network based on multi-information sources is proposed in this paper. Firstly, a fault diagnosis method based on multi-source information for distribution network is proposed from the point of view of the difficulty and the amount of difficulty in obtaining fault data. It includes the switch layer used to diagnose the simple fault quickly, the feeder layer to solve the complex fault under the condition of switch and protection, and the substation layer to accurately judge the multi-type fault in the complex system environment. At the same time, the dynamic jump strategy is adopted. According to the fault characteristics, the diagnosis entrance and structure are adjusted longitudinally. The method enhances the adaptability of each layer of diagnosis and improves the efficiency and accuracy of fault diagnosis. Secondly, from the point of view of actual distribution network system, a fault diagnosis method based on fuzzy Petri net and hierarchical multi-source information fusion is proposed. This method includes the switch layer of fault candidate set and the multi-source information fusion diagnostic layer based on directionally weighted fuzzy Petri net wavelet transform and improved D-S evidence theory fusion. This method can automatically adapt to the network topology change, and has good generality and fault tolerance. Finally, considering the accurate identification of distribution network fault types is the premise of fault location, after comprehensive analysis of distribution network fault characteristics, an immune neural network based distribution network fault type identification method is proposed. This method uses wavelet transform technology to extract fault features as the input of immune neural network, and then optimizes the weights of neural network by using the global search performance of immune algorithm, thus avoiding the neural network algorithm falling into local small points. This method can accurately realize the fault type identification under various fault modes, and has good robustness and convergence.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM727

【相似文献】