基于改进支持向量机的空气质量监测预警模型

发布时间：2018-03-27 14:26

本文选题：空气质量指数　切入点：参数优化　出处：《江苏大学学报(自然科学版)》2016年04期

【摘要】：在使用粒子群和组合预测方法改进传统支持向量机预测精度的基础上,构建了基于空气质量指数的城市空气质量监测预警模型.在参数优化方面,为了提高惩罚参数和核参数的选择精度,利用带收敛因子的粒子群算法,优化了网格搜索交叉验证法的参数筛选流程;在模型改进方面,为综合利用多种预测算法的优势,引入组合预测方式对灰色预测、时间序列预测和PSO-SVM模型的预测结果进行最优线性组合.结果表明:改进后的参数筛选流程和支持向量机的空气质量监测预警模型具有预测数据结构风险低、预测均方误差最小、运算精度高、运算速度快和适用性广等特征.
[Abstract]:On the basis of using particle swarm optimization and combination forecasting method to improve the prediction accuracy of traditional support vector machine, an urban air quality monitoring and warning model based on air quality index is constructed. In order to improve the accuracy of selecting penalty and kernel parameters, the particle swarm optimization algorithm with convergence factor is used to optimize the parameter selection flow of the cross-validation method for grid search, and to make comprehensive use of the advantages of various prediction algorithms in improving the model. The combination forecasting method is introduced to the grey forecast. The optimal linear combination of time series prediction and PSO-SVM model prediction results show that the improved parameter screening process and the air quality monitoring and warning model based on support vector machine have low risk of prediction data structure and minimum prediction mean square error. It is characterized by high precision, high speed and wide applicability.
【作者单位】：东北财经大学公共管理学院;东北财经大学国际商学院;东北财经大学统计学院;
【基金】：国家社会科学基金资助项目(14CRK019) 辽宁省社科规划基金资助项目(L14CTJ005) 辽宁省教育厅人文社科项目(W2014209;ZJ2013035) 辽宁省社科联辽宁经济社会发展立项课题(2016lslktzzx-01;2016lsljdwtzdian-02)
【分类号】：X831;TP18

【相似文献】