论文部分内容阅读
为了更好地预测室内热舒适度PMV指标,在分析模糊C-均值聚类方法与支持向量机方法的优势和互补性后,探讨了二者的结合方法,提出了一种基于模糊C-均值聚类预处理的支持向量机PMV指标预测系统。该方法把复杂的数据集看作多个群体的混合,每个群体采用单一的回归模型进行描述,使得大规模数据集的回归估计问题变成了一个多模型估计问题。将该系统应用于PMV指标预测中,与标准支持向量机方法相比,得到了较高的预测精度,从而说明了基于模糊C-均值聚类方法作为信息预处理的支持向量机学习系统的优越性。
In order to better predict the indoor thermal comfort PMV index, after analyzing the advantages and complements of the fuzzy C-means clustering method and the support vector machine method, this paper discusses the combination method of the two methods and proposes a fuzzy C-means Clustering Preprocessing Support Vector Machine PMV Indicator Prediction System. In this method, the complex data set is considered as a mixture of multiple groups. Each group is described by a single regression model, which makes the regression estimation problem of large-scale data set a multi-model estimation problem. Compared with the standard support vector machine method, the system has been applied to predict the PMV index, and higher prediction accuracy has been obtained, which shows the superiority of the system based on fuzzy C-means clustering as a support vector machine learning system for information preprocessing Sex.