摘要
提出智能优化支持向量机算法来提高模型的预测能力和泛化能力。该算法针对支持向量机噪声敏感问题采用小波方法对数据集去噪;利用核主成分分析方法提取数据特征;采用量子粒子群算法优化支持向量机超参数。将该优化算法应用于锅炉负荷短期预测,实验结果表明,该优化算法预测精度较高,收敛速度较快,泛化性能优于其他预测方法,且工程实现容易。
Intelligently optimal support vector machine （SVM） were introduced in electric utility boiler to improve short-term load forecasting accuracy and generalization ability. Wavelet transform is adopted to filter noise in training and testing data set. Kernel principle component analysis is used in feature selection. Then quantum-behaved particle swarm algorithm is chosen to determinate optimal hyper-parameter in SVM. This optimal algorithm has been tested on power plant and the results show that the prediction can get higher precision and convergence speed.
出处
《电子科技大学学报》
EI
CAS
CSCD
北大核心
2010年第2期316-320,共5页
Journal of University of Electronic Science and Technology of China
基金
贵州省自然科学基金(黔科合字20072004)
贵州省省长专项资金(黔省专合字(2007)14号)
关键词
预测
核主成分分析
优化
量子粒子群算法
支持向量机
forecasting
kernel principle component analysis
optimization
quantum-behaved particle swarm algorithm
support vector machines
