基于GA-PSO算法优化BP网络的短期电力负荷预测

发布时间：2018-05-19 05:31

本文选题：BP神经网络 + 短期电力负荷　；参考：《贵州师范大学》2014年硕士论文

【摘要】：随着电力市场的不断发展，电力负荷预测工作成为电力系统管理部门的一项重要工作。准确地进行电力负荷预测可以更好地制定电网规划方案以及发电机组的检修计划，可以更加合理地安排电网的运行方式。对于提高电力企业的经济效益和社会效益、保持电力系统的安全稳定运行、保障人们日常生活的有序进行具有重要的意义。本文首先介绍了电力负荷预测的研究背景、国内外研究现状以及研究意义，并且叙述了电力负荷预测的基本理论；其次，对现代预测关键技术进行了详细的介绍，，介绍了人工神经网络的基本理论，研究了BP神经网络的结构、BP神经网络的学习算法步骤及其优缺点，分析了遗传算法和粒子群优化算法的特点以及基本原理；再次，对BP神经网络预测模型进行了设计，主要是输入层和输出层的设计、隐含层的设计以及转移函数的确定；最后，分别采用遗传算法(GA)、粒子群优化算法(PSO)以及本文所提出的GA-PSO算法对BP神经网络的权值和阈值进行优化，分别建立了GA-BP神经网络预测模型、PSO-BP神经网络预测模型以及GA-PSO-BP神经网络预测模型。选取欧洲某地区的历史负荷数据、历史气温和日期类型等数据进行仿真实验，对该地区某一天24小时各整点时刻的负荷进行预测。并分析预测结果，比较各预测模型的性能。仿真实验结果表明GA-PSO-BP神经网络预测模型不仅加快了神经网络的收敛速度，而且提高了短期电力负荷的预测精度。
[Abstract]:With the development of power market, power load forecasting has become an important task in power system management department. Accurate load forecasting can better formulate the power network planning plan and maintenance plan of the generator set, and it can more reasonably arrange the operation mode of the power network. It is of great significance to improve the economic and social benefits of electric power enterprises, to maintain the safe and stable operation of power system and to ensure the orderly operation of people's daily life. This paper first introduces the research background of power load forecasting, the research status and significance at home and abroad, and describes the basic theory of power load forecasting. Secondly, the key technologies of modern forecasting are introduced in detail. This paper introduces the basic theory of artificial neural network, studies the learning algorithm steps of BP neural network and its advantages and disadvantages, analyzes the characteristics and basic principles of genetic algorithm and particle swarm optimization algorithm. The prediction model of BP neural network is designed, including the design of input layer and output layer, the design of hidden layer and the determination of transfer function. The weight and threshold of BP neural network are optimized by genetic algorithm (GA), particle swarm optimization (PSO) and GA-PSO algorithm proposed in this paper. The GA-BP neural network prediction model and the GA-PSO-BP neural network prediction model are established respectively. The data of historical load, historical temperature and date type of a certain region in Europe are selected to carry out simulation experiments to predict the load at each hour of 24 hours a day in that region. The prediction results are analyzed and the performance of each prediction model is compared. The simulation results show that the GA-PSO-BP neural network model not only accelerates the convergence rate of the neural network, but also improves the accuracy of short-term power load prediction.
【学位授予单位】：贵州师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM715

【参考文献】