详细信息
最小二乘支持向量机在API预测中的应用 被引量:1
Application of least square support vector machine in air pollution index prediction
文献类型:期刊文献
中文题名:最小二乘支持向量机在API预测中的应用
英文题名:Application of least square support vector machine in air pollution index prediction
作者:柳春[1];李四海[1];魏邦龙[2]
第一作者:柳春
机构:[1]甘肃中医学院信息工程学院,兰州730000;[2]兰州城市学院信息工程学院,兰州730070
第一机构:甘肃中医药大学信息工程学院(教育技术中心)
年份:2013
卷号:31
期号:4
起止页码:509
中文期刊名:沈阳师范大学学报(自然科学版)
外文期刊名:Journal of Shenyang Normal University:Natural Science Edition
收录:CSTPCD
基金:甘肃省教育厅高等学校研究生导师科研项目(0811-06)
语种:中文
中文关键词:最小二乘支持向量机;空气污染指数;气象因子;小波分解与重构;参数优化
外文关键词:LS-SVM; air pollution index; meteorological factors; wavelet decomposition and reconstruction; parameter optimization
摘要:传统的空气污染指数预测模型大多是以影响空气污染指数的重要气象因子作为输入,使用BP神经网络进行建模,模型的预测精度低且收敛速度慢。针对空气污染指数时间序列的非线性及多分辨率特性,提出了一种空气污染指数的最小二乘支持向量机预测模型。首先利用小波变换对原始的空气污染指数时间序列进行多尺度分解,以各尺度上的小波单支重构序列和重要的气象因子作为输入,然后使用该模型对兰州地区的空气污染指数进行了预测,最后讨论了模型参数的优化方法并使用网格法对两个参数进行了优化。仿真结果表明,与传统的BP神经网络预测模型相比,该模型具有更高的预测精度、更快的收敛速度及更好的稳定性。
The traditional air pollution index (API) prediction model is frequently based on the input of the key meteorological factors that affects the air pollution index and constructed by using the BP neural network, while the traditional model has lower prediction precision and slower convergence speed. In the light of the nonlinear and multi-- resolution characteristics of API time series, a new least square support vector machine(LS-SVM) prediction model was constructed. Discrete wavelet transform was used for the multiple-scaled decomposition of the original API time series, and the single wavelet reconstruction sequences and important meteorological factors were input. The model was applied to predict the API of Lanzhou, and the optimization of the parameters was discussed and the two parameters 9' and a was optimized hy the grid method. The result showed that, compared with the traditional BP neural network prediction model, the model had higher prediction precision, quicker convergence speed and better stability.
参考文献:
正在载入数据...