基于RBF神经网络的配电网线损计算与分析

发布时间：2018-03-14 23:10

本文选题：配电网线损　切入点：RBF神经网络　出处：《东北石油大学》2014年硕士论文　论文类型：学位论文

【摘要】：配电网线损是考核电力企业的一项主要经济指标,直接影响到电力部门的经济效益,所以降低配电网线损率就成为了电力部门迫切需要解决的问题。制定合理的降低电网损耗实施办法和线损考核指标就需要对电网线损进行准确计算和分析以便对电力部门的线损管理起到辅助和指导作用。本文首先对传统的配电网理论线损计算方法进行分析,针对配电网线路的结构和运行参数复杂、各个元器件的原始数据获取不全、自动化程度较低等特点,提出一种基于遗传算法的RBF神经网络配电网线损计算方法。另外,本文还引入了自适应遗传和种群更新策略对遗传算法进行改进,使RBF神经网络具有较高的收敛速度和精度,更能够满足实际要求。为了验证本文提出的计算方法更适用于配电网线损计算,我们以大庆油田电力集团所管辖的67条线路进行仿真计算,仿真结果证明了此算法具有计算精度高、网络模型简单、并拥有很高的适用性和推广性。另外,本文又将遗传优化的RBF神经网络与的RBF神经网络和BP网络进行对比,发现此算法无论在计算精度、记忆能力上都较其他算法有明显提高。最后在大庆油田电力集团的大力支持下开发了一套配电网线损计算软件,本软件的主要功能包括用户管理、数据库管理、配电网线损计算、线损分析、报表输出等功能。软件人机界面友好,执行效率较高,数据结构层次分明,数据访问效率高。另外,在软件开发后期经过现场工作人员的测试,保证了软件的完整性和可靠性。
[Abstract]:The distribution network line loss is one of the main economic indexes to assess the electric power enterprises, which directly affects the economic benefits of the power sector. Therefore, reducing the distribution network line loss rate has become an urgent problem to be solved by the power sector. It is necessary to accurately calculate and analyze the power network line losses in order to establish a reasonable way to reduce the network losses and assess the line losses. The line loss management of the electric power department plays an auxiliary and guiding role. In this paper, the traditional theoretical line loss calculation method of distribution network is analyzed, aiming at the characteristics of complex structure and operation parameters of distribution network, incomplete acquisition of original data of each component, low degree of automation, etc. This paper presents a genetic algorithm based on RBF neural network for line loss calculation. In addition, adaptive genetic and population updating strategies are introduced to improve the genetic algorithm, so that the RBF neural network has a higher convergence speed and accuracy. More able to meet the actual requirements. In order to verify that the proposed method is more suitable for the distribution network line loss calculation, we use 67 lines under the jurisdiction of Daqing Oilfield Electric Power Group to carry out the simulation calculation. The simulation results show that the algorithm has a high accuracy and a simple network model. In addition, the RBF neural network based on genetic optimization is compared with the RBF neural network and BP neural network, and it is found that the algorithm is better than other algorithms in computing accuracy and memory ability. Finally, with the support of Daqing Oilfield Electric Power Group, a set of software is developed. The main functions of the software include user management, database management, distribution network line loss calculation, line loss analysis. The software has friendly man-machine interface, high execution efficiency, clear data structure and high data access efficiency. In addition, the integrity and reliability of the software are ensured by field staff tests in the later stage of software development.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP183;TM744

【共引文献】