|
| 摘要: |
| 考虑气象因素对负荷的影响，提出了一种模糊支持向量机SVM(Support Vector Machine)的短期负荷预测方法。首先选取预测日前4星期中差异评价函数小于给定经验值的已知日作为相似日学习样本．然后利用隶属度函数对影响负荷特征因素向量的分量进行模糊处理，得到SVM的训练样本集．拟合负荷和影响因素之间的非线性关系。对24点每点建立一个SVM预测模型，采用改进的序列极小优化算法实现对SVM的快速训练。算例数据包括每天的气象数据和24点负荷数据．以最大相对误差和平均误差评价预测结果，表明所提方法简便快速且实用有效。 |
| 关键词: 短期负荷预测 支持向量机 核函数 隶属度函数 |
| DOI： |
| 分类号:TM715 |
| 基金项目: |
|
| Short-term load forecast based on fuzzy support vector machine method |
|
GONG Deng-cai LI Xun-ming LI Lin-feng
|
| Abstract: |
| Considering the influence of weather on loads,a method based on fuzzy SVM(Support Vector Machine) is presented for the short,term load forecast. The days during last four weeks,whose difference estimation function values are smaller than the given experiential value,are selected as similar days for learning. The influencing factors of eigenvector are processed by fuzzy membership functions to form the training sample set for SVM and fit the nonlinear relationship between loads and influencing factors. SVM forecast models are established for every point of 24 ,point loads,and an improved sequential minimal optimization method is used to train SVM. The example provides daily weather data and 24,point load data,and the maximum relative error and mean error are selected to evaluate the forecast. Results show the proposed method simple,fast,practical and effective. |
| Key words: short,term load forecast,support vector machine,kernel function,membership function, |
