基于BP神经网络的电力负荷预测模型设计与系统实现

发布时间：2018-09-08 13:03

【摘要】：随着近年来我国国民经济的快速发展,各个经济部门对于电力能源的需求量也越来越大,电力管理部门面临着重要的挑战。由于电力资源具有随产随用的特殊性质,所以在电能的生产过程中必须按照实际的需求量进行按需生产,因此如何对未来的电力负荷进行准确预测已经成为电力系统调度管理部门的重要挑战。在以往的电力负荷预测工作中主要采用的是线性假设预测技术,面对预测精度日益提高的需求已经无法满足电力管理部门的要求,所以如何提高电力负荷预测的准确度是未来工作的重点。随着神经网络技术的快速发展,其自适应能力强、容错性好等优点已经成为电力负荷预测模型的重要基础技术,并得到了广泛的认可与关注。本文通过对电力负荷预测基本理论进行分析,并结合神经网络技术以及小波阈值去噪技术构建了电力负荷预测模型,同时使用.NET开发平台实现了一个电力负荷预测软件平台,具体而言,论文的研究工作主要包括:(1)通过对常见的电力负荷预测算法进行分析,同时结合电力负荷预测中的非线性特征设计了基于神经网络技术的负荷预测模型,对神经网络的输入层、隐含层以及结果输出层进行设计,完成了系统的核心预测算法模型设计。(2)对于电力负荷预测过程中使用的历史负荷数据中存在的干扰因素过多以及数据噪声较大等特征,提出了采用小波阈值去噪的方式对数据中的噪声进行去除处理,同时对传统的小波阈值去噪算法进行了一定的改进操作。(3)在理论分析的基础上,完成了电力负荷预测软件系统的业务建模分析,提出了系统所需实现的各项功能,并通过系统功能结构设计、对象模型设计、负荷预测模型设计和数据库设计为系统的功能实现提供基础。(4)按照系统研发过程中以及论文研究工作中存在的问题进行了总结与分析,并针对当前系统中仍然存在的问题与不足提出了相关改进与完善意见,作为后续研究工作的指导。
[Abstract]:With the rapid development of China's national economy in recent years, the demand for power energy in various economic sectors is also increasing, and the power management department is facing an important challenge. Due to the special nature of power resources, the production of electric energy must be carried out according to the actual demand. Therefore, how to accurately predict the future power load has become an important challenge for the power system dispatching management department. In the previous work of power load forecasting, the linear hypothesis forecasting technology is mainly used. In the face of the increasing demand of forecasting accuracy, it can not meet the requirements of power management departments. Therefore, how to improve the accuracy of power load forecasting is the focus of future work. With the rapid development of neural network technology, its advantages such as strong adaptive ability and good fault tolerance have become an important basic technology of power load forecasting model, and have been widely recognized and paid attention to. In this paper, the basic theory of power load forecasting is analyzed, combined with neural network technology and wavelet threshold de-noising technology, the power load forecasting model is constructed. At the same time, a power load forecasting software platform is implemented by using .NET development platform. Specifically, the research work includes: (1) through the analysis of common power load forecasting algorithms, combined with the nonlinear characteristics of power load forecasting, a load forecasting model based on neural network technology is designed. The input layer, hidden layer and result output layer of neural network are designed. The core prediction algorithm model of the system is designed. (2) for the characteristics of too many interference factors and large data noise in the historical load data used in the power load forecasting process, The wavelet threshold denoising method is proposed to remove the noise in the data, and the traditional wavelet threshold denoising algorithm is improved. (3) on the basis of theoretical analysis, The business modeling and analysis of the power load forecasting software system is completed, and the functions that the system needs to realize are put forward, and the system function structure design and object model design are given. Load forecasting model design and database design provide the basis for the realization of system functions. (4) according to the existing problems in the research and development of the system and in the research work of the paper, the paper summarizes and analyzes. In view of the existing problems and deficiencies in the current system, the paper puts forward some suggestions for improvement and perfection, which can be used as the guidance of the following research work.
【学位授予单位】：西华大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM715;TP183

【相似文献】