基于人工蜂群优化LS-SVM短期负荷预测研究

发布时间：2018-02-16 13:30

本文关键词： 短期负荷预测最小二乘支持向量机人工蜂群算法混沌搜索　出处：《辽宁工程技术大学》2014年硕士论文　论文类型：学位论文

【摘要】：短期负荷预测是电力部门非常重要的一项工作,它关系到发电单位对发电计划的合理安排,所以对其进行研究是非常必要的。短期负荷预测的结果是否精准影响着电力部门的计划交易量和电力系统是否能安全稳定地运行。因此,采用适当的短期负荷预测方法和最大程度地提高预测精度和准确性已成为负荷预测领域的重要研究课题。首先,本文介绍了负荷预测的研究背景及意义并对负荷预测进行了分类,重点阐述了短期电力负荷预测的特点、方法和研究现状等；其次,结合电力负荷的特性,提出建立基于最小二乘支持向量机的预测模型。由于模型参数对预测的结果有着很大的影响,所以最小二乘向量机参数的优化非常重要,本文分别采取人工蜂群算法和改进蜂群算法对参数进行寻优,对两种算法的原理和流程进行了说明,并对两种算法的优化性能进行验证。最后分别建立基于LS-SVM、ABC-LSSVM和CABC-LSSVM的预测模型,针对辽宁某地区的负荷进行预测,并对比各模型的预测结果,结合影响负荷预测的因素进行分析并得出结论,预测结果验证了本文人工蜂群算法优化是有效的,其中基于改进蜂群优化LSSVM预测精度最高。
[Abstract]:Short-term load forecasting is a very important task in the power sector. It is related to the reasonable arrangement of power generation plans by power generation units. So it is very necessary to study it. Whether the result of short-term load forecasting affects the planned trading volume of the power sector and whether the power system can operate safely and stably. It has become an important research topic in the field of load forecasting to adopt appropriate short-term load forecasting methods and to improve forecasting accuracy and accuracy to the greatest extent. This paper introduces the research background and significance of load forecasting, and classifies load forecasting, focusing on the characteristics, methods and research status of short-term power load forecasting. Secondly, combined with the characteristics of power load, A prediction model based on least squares support vector machine (LS-SVM) is proposed. Because the parameters of the model have a great influence on the prediction results, it is very important to optimize the parameters of LS-SVM. In this paper, artificial bee colony algorithm and improved bee colony algorithm are adopted to optimize the parameters, the principle and flow of the two algorithms are explained, and the optimization performance of the two algorithms is verified. Finally, the prediction models based on LS-SVMMA-ABC-LSSVM and CABC-LSSVM are established, respectively. According to the load forecast of a certain area in Liaoning Province, the forecasting results of each model are compared, and the factors affecting load forecasting are analyzed and concluded. The prediction results verify the effectiveness of the artificial bee colony algorithm optimization in this paper. The prediction accuracy of LSSVM based on improved colony optimization is the highest.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM715

【参考文献】