基于负荷规律性分析的空间负荷预测方法研究

发布时间：2018-04-30 14:48

本文选题：空间负荷预测 + 负荷规律性　；参考：《东北电力大学》2017年硕士论文

【摘要】：空间负荷预测(spatial load forecasting,SLF)是电力系统规划设计的先决条件,因此,高精度的空间负荷预测结果对城市电网的规划设计具有极其重要的意义。与系统的负荷预测相比,空间电力负荷具有明显的时空分布特性。历史数据是SLF的根基,预测时所使用数据的真实性及准确性将影响预测精度的高低。当电力系统由于通信等原因而出现人为干扰因素时,历史负荷数据中会夹杂许多可疑数据的情况。这些可疑数据的出现将可能导致预测模型和结果与实际水平间的差异超出系统的阈值,从而使预测工作失去实际意义。因此,欲提高预测结果的可信度,对预测基础数据进行分析处理就显得非常重要。选取恰当合理的历史负荷进行合理性分析,去伪存真,准确、经济、合理地进行电网变电站布点、线路走廊规划,使SLF在城市电网规划中发挥更大的作用。本文以大量历史负荷数据为基础,着重分析了城市配电网电力负荷的变化特点,充分挖掘了历史负荷数据中蕴含的变化规律性,提取出历史负荷数据本身含有的趋势性和规律性分量,剥离对负荷预测带来不利影响的随机性分量,确定不含内蕴随机分量的元胞负荷的合理最大值,实现了较为精确的负荷预测。研究负荷周期性分析理论,结合集合经验模态分解(Ensemble Empirical Mode Decomposition,EEMD)理论,提出了空间负荷预测中基于EEMD分解来确定元胞合理最大值的方法,用以避免元胞负荷实测数据将测量、通信等过程中的误差带入预测过程,而导致预测结果精度降低的问题。通过EEMD分解将各元胞负荷分解成一系列本征模函数,并建立滤取机制,分别重构表征规律性部分的主体分量和表征随机性部分的随机分量。剔除对预测结果带来不利影响的随机误差,将剩余部分最大值作为元胞负荷的合理最大值。针对城市电网总量负荷预测,提出一种城市电网总量负荷的双向预测方法。利用用电量与电力负荷之间的相关关系,将历史用电量数据转化为电力负荷数据,并采用双向预测的方法进行预测。该方法充分挖掘历史用电量数据与电力负荷数据之间的内在联系,以历史用电量数据为基础求得电力负荷数据,从而丰富负荷数据结构,避免了历史数据的不充分和直接在原始电力负荷年最大值的历史数据的基础上进行预测的缺陷,提高了数据的准确性与稳定性,降低了电力负荷年最大值数据的波动性对预测带来的不利影响。将元胞负荷合理最大值的确定方法引入到空间负荷预测中,在网格化负荷密度指标法的基础上,通过分析不同分辨率下对预测结果的误差趋势,以及不同分辨率下的计算速度,选取最佳的空间分辨率以提高空间负荷预测的准确性。
[Abstract]:Spatial load forecasting (spatial load forecasting) is a prerequisite for power system planning and design. Therefore, high precision spatial load forecasting results are of great significance for urban power network planning and design. Compared with the load forecasting of the system, the spatial electric load has obvious spatial and temporal distribution characteristics. Historical data is the foundation of SLF. The accuracy and authenticity of the data used in prediction will affect the accuracy of prediction. When artificial interference occurs in power system due to communication and other reasons, the historical load data will be mixed with many suspicious data. The appearance of these suspicious data may cause the difference between the prediction model and the actual level to exceed the threshold of the system, so that the prediction work will lose its practical significance. Therefore, in order to improve the reliability of prediction results, it is very important to analyze and process the basic prediction data. Selecting appropriate and reasonable historical load for rationality analysis, distinguishing false things, retaining real things, accurately, economically and reasonably carrying out substation distribution and line corridor planning in power grid makes SLF play a more important role in urban power network planning. Based on a large number of historical load data, this paper emphatically analyzes the characteristics of power load variation in urban distribution network, and fully excavates the variation regularity contained in historical load data. The trend and regularity components contained in the historical load data are extracted, the random components that bring adverse effects on load forecasting are stripped, and the reasonable maximum value of the cellular load without intrinsic random components is determined. More accurate load forecasting is realized. Based on the theory of periodic load analysis and the theory of Ensemble Empirical Mode decomposition (EEMD), a method based on EEMD decomposition to determine the reasonable maximum value of cell in spatial load forecasting is proposed, which is used to avoid the measurement of the measured data of cell load. The error in the process of communication is brought into the prediction process, which leads to the decrease of the precision of the prediction result. The cellular load is decomposed into a series of eigenmode functions by EEMD decomposition, and the filtering mechanism is established to reconstruct the principal component representing the regular part and the random component representing the random part, respectively. The maximum value of the remaining part is taken as the reasonable maximum value of the cell load. This paper presents a bidirectional forecasting method for the total load of urban power network. Based on the correlation between electricity consumption and power load, the historical power consumption data is transformed into power load data, and the bidirectional forecasting method is used to forecast the data. This method fully excavates the internal relation between the historical electricity consumption data and the power load data, and obtains the electric power load data based on the historical electricity consumption data, thus enriches the load data structure. It avoids the insufficiency of historical data and the defect of forecasting directly on the basis of historical data of the original annual maximum value of electric load, and improves the accuracy and stability of the data. The volatility of annual maximum data of power load is reduced to the adverse effect of forecasting. The method of determining the reasonable maximum value of cellular load is introduced into spatial load forecasting. On the basis of grid load density index method, the error trend of prediction results under different resolutions and the calculation speed at different resolutions are analyzed. The best spatial resolution is chosen to improve the accuracy of spatial load forecasting.
【学位授予单位】：东北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM715

【参考文献】