智能配电网馈线负荷预测系统研究

发布时间：2018-06-12 00:20

本文选题：智能配电网 + 馈线　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：智能配电网馈线负荷预测，可以为配电网电能调度提供决策信息，指导用户智能用电，平抑负荷波动，对保证配电网安全、经济的运行具有重要意义。随着我国配电网的小型分布式电源如风电、光伏等应用发展迅速，其所发电能受到风速、日照强度等不确定因素的影响，具有显著的随机间歇性。这间歇性加剧配电网馈线上的负荷波动，干扰配电网的安全稳定运行，对智能配电网馈线负荷预测带来新的挑战。针对智能配电网馈线负荷与传统馈线负荷具有不同的特点，按照负荷、发电来源将馈线负荷划分为用户负荷、分布式风电负荷和光伏负荷等部分。考虑分布式负荷的波动对馈线负荷预测的影响，参照馈线潮流方向将分布式负荷定性为负负荷，并定义了馈线净负荷的概念。在负荷划分的基础上，对智能配电网馈线上的各类负荷进行特征分析，确定各类负荷预测所需的数据类型，，分别采用C-模糊聚类与K-均值聚类的方法对用户负荷和光伏负荷进行聚类分析，建立各类馈线负荷模式。根据各类馈线负荷的特性，分别提出合适的各类馈线负荷预测方法。建立基于鲁棒回归和改进的Elman神经网络的馈线用户负荷预测模型。建立基于小波去噪的ARMA时间序列的馈线风电负荷预测模型。建立基于GRNN神经网络的馈线光伏负荷预测模型。确定基于馈线负荷重组的馈线净负荷预测方法。根据平均相对误差和VAR值的误差评价指标，评价所提出方法的预测精度，分析预测误差风险。根据预测所需数据类型及其相互关系，设计预测系统后台数据仓库，建立基于数据挖掘的智能配电网馈线负荷预测系统。该系统具有负荷浏览分析、建立各类负荷模式、各类馈线负荷预测、馈线净负荷预测和预测误差分析等主要功能。最后以某含分布式电源的智能配电网馈线为例，根据该网历史负荷数据，实现系统各功能，验证所设计预测系统的有效性与实用性。本文工作得到国家电网公司科技项目《智能配电网控制运行模拟仿真关键技术研究》(DZB17201200260)的资助。
[Abstract]:The intelligent distribution network feeder load forecasting can provide decision information for distribution network power dispatching, guide users to use electricity intelligently, suppress load fluctuation, and is of great significance to ensure distribution network safety and economic operation. With the rapid development of small distributed generation (DG) such as wind power and photovoltaic (PV) in China's distribution network, the power generation can be affected by uncertain factors such as wind speed, sunshine intensity and so on. This intermittency intensifies the load fluctuation on the feeder line of the distribution network, interferes with the safe and stable operation of the distribution network, and brings a new challenge to the intelligent distribution network feeder load forecasting, aiming at the different characteristics between the intelligent distribution network feeder load and the traditional feeder load. According to the load, the feeder load is divided into user load, distributed wind load and photovoltaic load. Considering the influence of distributed load fluctuation on feeder load forecasting, the distributed load is defined as negative load according to the direction of feeder power flow, and the concept of net load of feeder line is defined. Based on the characteristic analysis of various kinds of loads on the feeder line of intelligent distribution network, the data types needed for load forecasting are determined, and the methods of C- fuzzy clustering and K-means clustering are used to cluster the user load and photovoltaic load, respectively. According to the characteristics of various kinds of feeder load, the appropriate load forecasting methods are put forward. A feeder user load forecasting model based on robust regression and improved Elman neural network is established. An ARMA time series model for wind power load forecasting based on wavelet denoising is established. A feed-line photovoltaic load forecasting model based on GRNN neural network is established. The net load forecasting method based on feeder load recombination is determined. According to the average relative error and the error evaluation index of VAR value, the prediction accuracy of the proposed method is evaluated, and the prediction error risk is analyzed. According to the data types and their relationships, the backstage data warehouse of the forecasting system is designed. The intelligent distribution network feeder load forecasting system based on data mining is established. The system has the main functions of load browsing analysis, setting up all kinds of load modes, various kinds of feeder load forecasting, feeder net load forecasting and forecasting error analysis and so on. Finally, taking the feeder of a smart distribution network with distributed power as an example, according to the historical load data of the network, the functions of the system are realized. To verify the effectiveness and practicability of the designed prediction system, this paper is supported by the State Grid Corporation's Science and Technology Project "key Technology Research on Simulation of Intelligent Distribution Network Control Operation Simulation" (DZB17201200260).
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM76

【参考文献】