基于最小二乘支持向量机改进算法的短期负荷预测-电力系统及其自动化专业论文.docxVIP

东北电力大学硕士学位论文Abstract 东北电力大学硕士学位论文 Abstract Short-term load forecasting is one of the important contents of power system operating and dispatching，and it is the basis of making electricity purchase plans and economic operation．So short-term load forecasting has great significance in power system security．With the development of power system，the load forecasting methods draw more and more attention，which have hi曲precision and Can considerate different influencing factors．According to nonlinear characteristics of the factors affecting accuracy of Load Prediction，a new method of short—load forecasting based on LSSVM Can enhance prediction precision and speed．The new algodthrn has important theory research signification and utility value． The paper introduces the principle of LSSVM which is applied to short—load forecasting of power system．However,when the sample size is large，the training speed is slow relatively．Considering this，LSSVM incremental training algorithm is applied to increase the training speed．LSSVM losses the sparseness of solution compared with SVM，which affects accuracy of load prediction．Therefore，an improvement of pruning algorithm is introduced to impose sparseness upon the solution of LSSVM incremental learning algorithm． According to the actual need of load forecasting，a set forecasting evolution system is developed，which is applied in Yichun，Jiagedaqi，and Harbin power grids of Heilongj iang province．Moreover,the system provides more accurate and helpful forecasting results for the operators． Key words：，power system；short-load forecasting；SVM；LSSVM； incremental training；pruning algorithm ．Ⅱ- 论文原创性声明 论文原创性声明 本人声明，所呈交的学位论文系在导师指导下本人独立完成的研究成果。文 中依法引用他人的成果，均己做出明确标注或得到许可。论文内容未包含法律意 义上已属于他人的任何形式的研究成果，也不包含本人已用于其他学位申请的论 文或成果。 本人如违反上述声明，愿意承担以下责任和后果： 1． 交回学校授予的学位证书： 2． 学校可在相关媒体上对作者本人的行为进行通报； 3． 本人按照学校规定的方式，对因不当取得学位给学校造成的名誉损害， 进行公开道歉； 4． 本人负责因论文成果不实产生的法律纠纷。 论文作者签名： 附录5 论文知识产权权属声明 本人在导师指导下所完成的论文及相关的职务作品，知识产权归属东北电力 大学。学校享有以任何方式发表、复制、公开阅览、借阅以及申请专利等权利。 本人离校后发表或使