导　　师： 文福拴

学科专业： 080802

授予学位： 硕士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 电力系统负荷预测是电力系统规划、计划、调度、营销以及市场交易等部门决策时的重要依据。而且随着智能电网和电动汽车等的发展，负荷预测的重要性尤其突显。因此，提高负荷预测的准确率具有十分重要的意义。 电力负荷预测依赖于大量历史数据及相关因素分析，预测结果在很大程度上依赖于所收集到的历史数据的可靠性与相关因素分析资料的详细准确度。在此背景下，本论文主要从以下两个方面展开研究，以提高负荷预测的准确率。 其一，对收集到的历史数据，主要研究负荷数据的处理以提高其可靠性。因为来自SCADA系统数据库的电力负荷数据一般会有些异常数据，通过同时考虑负荷的横向连续性和纵向连续性，把负荷数据按照日期排列成二维数据集，首先采用基于密度的方法，在两个维度中对负荷数据中由于信道错误、极大极小负荷以及突发事故等原因产生的异常数据，进行整体上地辨识与修正；然后在此基础上，对负荷数据中可能存在的冲击负荷和由于信道噪声引起负荷锯齿状波动的毛刺负荷等，采用基于二维小波阈值去噪的方法，在两个维度中对负荷数据进行去噪处理，进一步提高负荷数据的质量。 其二，对相关因素的分析资料方面，主要研究了年最大降温负荷的测算与分析、负荷与温度之间的关联性分析与分行业负荷的构成分析等。以期这些分析能够给人工负荷预测专家根据实际情况修正负荷预测结果，对提高负荷预测的准确率提供一定的辅助作用。 Load forecasting is one of the most important issues in power system planning, dispatching andmarket operation, and it becomes more important with the development of the smart-grid and plug-inhybrid electrical vehic le /(PHEV/). Therefore, it is of great significance to improve the accuracy of loadforecasting. Load forecasting is heavily dependent on historial data and related factors. It is the objective of thisthesis to do systematical research as as to improve the accuracy of load forecasting. On the one hand, as to the historical load data, there are usually some abnormal data in the electricload database derived from the SCADA system, and the accuracy of load forecasting could then beimpacted. Hence, it is necessary to identify and then correct the abnormal data before the load data areused for load forecasting or power system analysis. However, so far the existing research work in thisarea is mainly done in one dimension. Given this background, considering both horizontal and verticalcontinuities of electric loads, firstly, a density evaluation based method is presented to identify and thencorrect abnormal data in two dimensions. Then, a new method for load de-noising is presented based onthe two-dimension wavelet threshold de-noising. Specifically, the load data is transformed into a matrixof gray-scale images by normalization. The images are processed by employing the two-dimensionwavelet threshold de-noising method. Finally, the de-noised data are obtained after de-normalization.The feasibility and efficiency of the developed method are demonstrated by the improvement of loadforecasting accuracy. On the other hand, as to the analyzed material of related factor, an effort is mainly made to calculateand analyze the annual maximum high-temperature related loads by employing the maximum load-basedcomparison method and the base load-based comparison method respectively, to do the study of thecorrelation analysis and sensitivity analysis between meteorological load and temperature by dividing theload into meteorological load and economic load, and to calculate the load composition of the wholepower system by employing the statistical inference method based on the monthly electricityconsumption data as well as the load data of each kind of industry loads collected by the load control system. And the result shows that these works could provide efficient decision-making assistance forpower system planning, dispatching, market operation, as well as accurate load forecasting.

关 键 词： 电力系统 负荷数据处理 负荷相关因素分析 密度估计 二维小波阈值去噪

分 类 号： [TM714]

领　　域： [电气工程]