一种基于支持向量回归的混合建模方法

发布时间：2018-02-16 02:42

本文关键词： 混合建模支持向量回归(SVR) D-Markov机小波数据驱动的建模　出处：《北京航空航天大学学报》2017年02期 　论文类型：期刊论文

【摘要】：近年来,随着计算能力的不断提高,数据驱动的建模方法受到了广泛的关注,对单模式系统进行定量分析的建模方法获得了诸多研究。然而,实际应用中大多数系统为多模式系统,不但各个模式有着不同的连续行为,连续状态还会在模式之间进行切换。针对这一情形,本文提出了经验概率混合自动机模型,并提出了针对该模型的基于支持向量回归(SVR)的多模式定性定量混合建模方法。该方法使用小波技术识别模式切换点,并在各个模式下单独建立支持向量模型,最后使用D-Markov机整合模型。经实例验证,该方法与传统支持向量回归模型的稳定性接近,但精确程度显著提高。
[Abstract]:In recent years, with the continuous improvement of computing power, data-driven modeling methods have received extensive attention. Modeling methods for quantitative analysis of single-mode systems have been studied. In practical applications, most systems are multi-mode systems, which not only have different continuous behaviors, but also switch between patterns. In this paper, an empirical probabilistic hybrid automaton model is proposed. A multi-pattern qualitative and quantitative hybrid modeling method based on support vector regression (SVR) for this model is proposed. The wavelet technique is used to identify pattern switching points, and support vector models are established separately in each mode. Finally, D-Markov machine integration model is used. It is proved by an example that the stability of this method is close to that of the traditional support vector regression model, but the accuracy is improved significantly.
【作者单位】：北京航空航天大学电子信息工程学院;西安卫星测控中心宇航动力学国家重点实验室;
【基金】：国家“973”计划~~
【分类号】：TP18

【参考文献】