电力系统短期负荷预测的分析与研究

发布时间：2018-06-20 17:13

本文选题：电力系统负荷预测 + 支持向量机　；参考：《西安理工大学》2017年硕士论文

【摘要】：在电力体制改革的新形势下,电价竞争机制介入电力市场,市场参与和系统运营人员对短期负荷预测越来越看重。快速准确的负荷预测有利于进行适当的计划电力交易、制定适当的操作计划和投标策略。随着新理论和新技术的发展,对负荷预测新方法的研究仍在不断地深入进行。支持向量机是一种数据预测的常用方法,与其它方法相比,表现出了较好的性能,能较好的实现结构风险最小化思想,可应用于模式识别和处理回归问题等诸多领域。论文利用支持向量机在非线性学习和预测性能上的优点,针对短期电力负荷预测的各种影响因素的非线性特性,支持向量机在参数选择上比较困难且传统方法存在一些缺陷问题;以及不同的核函数性能一般不同,在支持向量机的核函数选择上仍存在的一些问题,提出基于支持向量机的电力系统短期负荷预测的优化方法。分别针对双线性搜索法、网格搜索法的存在问题,提出一种改进的双线性搜索法和改进网格搜索法,通过实验对比,验证了改进方法的预测精度和时效性。此外,针对混合核函数权重系数难以准确选取的问题,采用粒子群优化实现权重的优化求解;针对粒子群算法存在的容易陷入局部最优的问题,在迭代过程中,根据适应度值动态调整惯性权重、局部加速常数及全局加速常数变异粒子,提出了一种改进的粒子群算法,实验结果表明改进粒子群混合核函数比单核核函数有更高的预测精度。
[Abstract]:In the new situation of electric power system reform, the electricity price competition mechanism is involved in the electricity market, market participation and system operators pay more and more attention to short-term load forecasting. Fast and accurate load forecasting is helpful for proper planned electricity transaction, proper operation plan and bidding strategy. With the development of new theory and new technology, the research of new load forecasting methods is still going on. Support vector machine (SVM) is a common method for data prediction. Compared with other methods, it has better performance and can realize structural risk minimization. It can be used in many fields such as pattern recognition and regression problems. Based on the advantages of support vector machine (SVM) in nonlinear learning and forecasting performance, this paper aims at the nonlinear characteristics of various factors affecting short-term power load forecasting. Support vector machine (SVM) is difficult to select parameters and has some defects in traditional methods. An optimization method for short term load forecasting of power system based on support vector machine (SVM) is proposed. In order to solve the problems of bilinear search and mesh search, an improved bilinear search method and an improved grid search method are proposed. The prediction accuracy and timeliness of the improved method are verified by experimental comparison. In addition, to solve the problem that the weight coefficient of hybrid kernel function is difficult to select accurately, particle swarm optimization is used to solve the weight optimization. Based on the dynamic adjustment of inertia weight, local acceleration constant and global acceleration constant, an improved particle swarm optimization algorithm is proposed. The experimental results show that the improved particle swarm hybrid kernel function has higher prediction accuracy than the single kernel function.
【学位授予单位】：西安理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM715

【参考文献】