基于RELM的时间序列数据加权集成分类方法

A weighted ensemble classification method for time series data based on regularized extreme learning machine

时间序列数据通常是指一系列带有时间间隔的实值型数据,广泛存在于煤矿、金融和医疗等领域。为解决现有时间序列数据分类问题中存在的含有大量噪声、预测精度低和泛化性能差的问题,提出了一种基于正则化极限学习机(RELM)的时间序列数据加权集成分类方法。首先,针对时间序列数据中所含有的噪声,利用小波包变换方法对时间序列数据进行去噪处理。其次,针对时间序列数据分类方法预测精度低、泛化性能较差的问题,提出了一种基于RELM的加权集成分类方法。该方法通过训练正则化极限学习机(RELM)隐藏层节点数量的方法,有效选取RELM基分类器;通过粒子群优化(PSO)算法,对RELM基分类器的权值进行优化;实现对时间序列数据的加权集成分类。实验结果表明,该分类方法能够对时间序列数据进行有效分类,并提升了分类精度。

Time series data usually refer to a series of real value data with time interval,which widely exists in coal mine,finance,medical and other fields.In order to solve the problems of large amount of noise,low prediction accuracy and poor generalization performance in the existing time series data classification problems,a weighted ensemble classification method based on regularized extreme learning machine(RELM)is proposed.Firstly,aiming at the noise contained in time series data,the wavelet packet method is used to denoise time series data.Secondly,in view of the low prediction accuracy and poor generalization performance of time series data classification method,a weighted ensemble classification method based on RELM is proposed.By training the number of hidden layer nodes of RELM,RELM base classifier is effectively selected.Through PSO method,the weight of RELM based classifier is optimized.Finally,weighted ensemble classification is performed on the time series data.Experimental results show that the method can effectively classify time series data and improve the classification accuracy.