导　　师： 聂一雄

学科专业： 080802

授予学位： 硕士

作　　者： ;

机构地区： 广东工业大学

摘　　要： 电网电能损耗是供电企业的一项重要经济技术指标,也是电网规划、生产技术管理、电网运行和经营管理水平的综合反映。面对我国电力紧缺的情况,最有效的办法就是“节能”和“开源”。在国家大力提倡优化结构、提高效益、节能降耗和污染减排的今天,节能降耗显得尤其重要。网/(线/)损率是供电企业管理水平的综合反映,不断降低电能损耗,是企业提高经济效益的必然要求。做好理论线损计算工作,不但有利于科学规划电网建设,而且有助于合理配置无功补偿设备,强化变压器运行管理与线损管理,降低电网损耗,对搞好节能降耗和提高电力企业经济效益具有非常重要的意义。 本文首先对配电网理论线损计算方法及其应用研究选题背景及研究意义进行介绍,阐述了本课题研究的必要性。通过对比分析现有线损分析计算方法的异同,研究多种因素对线损计算的影响,本文将人工神经网络理论引入到线损计算以综合智能地考虑影响线损计算的随机因素,以达到配网理论线损快速计算,并提高计算准确率。结合配电网理论线损计算的新方法应用研究,选择基于RBF、BP算法的神经网络作为本文的计算分析方法。 人工神经元网络模型算法不同于传统计算方法,它不需要确定计算结果与自变量的直接关系或回归方程。RBF及BP模型理论上可映射任意复杂的非线性关系,固而适于模拟线损与特征参数间的非线性关系,而且由于其是面向数据的,因而对任何配电网都适合。考虑到神经网络模型的一些固有特性以及线损与特征参数间存在的关系,为提高网络学习精度与效率,本文提出了一种简单而有效的数据处理法,首先将样本数据归一化处理后用神经网络模型对其训练,然后再跟另一组数据进行检验对比。本文研究过程,首先是用RBF网络训练并计算线损,根据结果寻求最优网络参数以得最佳计算结果,然后用BP网络训练并计算线损,改变BP网络训练次数及算法,直到输出结果误差最小,最后是选取两种网络计算的最优结果,模拟一个典型日负荷曲线图,从而了解在该日负荷曲线下神经网络所算的网损耗电量和实际网损耗电量的误差,证实神经网络运用在配电网理论线损计算的可行性。 Energy loss is an important economic and technical index of power grid, also a comprehensive reflection of network planning, technology management, network operation and management level. Facing our country's power shortages situation, the most effective way is to "save energy" and "find new energy". At the national advocate of optimizing structure, improving efficiency, energy saving and pollution reduction today, Energy-saving and consumption reduction is extremely important. The line losses' rate is a comprehensive reflection of the level of management, try to cut the line losses is a requirement for enterprise to improve its economic efficiency. Doing well the line losses calculation work, not only conducive to scientific planning of power grid construction, but also helpful to allocate reactive power compensation equipment reasonable, strengthen the operation management of transformer and line losses, reduce line losses, and has a very important significance for improving enterprise efficiency and energy-saving. In the beginning, this paper introduces the background and significance of theoretical energy loss research in distribution network. Then through comparative analysis the differences of the existing line losses calculation methods, explains the various factors on the calculation. Then, in order to achieve the goal of fast calculation and high calculation accuracy, this paper introduces ANN to the line losses calculation to consider the factors intelligently. Finally, combines with the new calculation method research of energy loss in distribution network, this paper selects neural network based on RBF, BP as a method of calculation. The method of ANN is different from the traditional method, it does not need the direct relationship or the regression equation of results and variables. RBF and BP model can map every complex nonlinear relationship in theory, so it is suitable to simulate the nonlinear relationship of line losses and power grid parameters, and because it is data-oriented, it can be applied to any distribution network. On account of the characteristics of neural network and the relationship between line losses and power grid parameters, in order to improve accuracy and efficiency of neural network learning, this paper presents a simple and efficient data processing method:training the neural network after normalized all the sample data, and then take calculation results compare with another group of test data. The research process:first, train the RBF network and calculate the line losses, according to the results, find the optimal parameters of the network. Then train BP network and calculate line losses, changes the training epochs and the modeling algorithm, find the optimal output. Finally select the optimal results to simulate a typical daily load curve, fine the error of electricity quantity, and confirm that neural network using in line losses calculation of distribution network is feasible.

关 键 词： 节能降耗 配电网 理论线损 网络 网络

分 类 号： [TM744]

领　　域： [电气工程]