数据驱动下基于时间序列云模型的特征选择聚类算法研究OACHSSCD

Due to the multivariate and high-dimensional nature of time series data,the difficulty of extracting important fea-tures increases,thereby reducing the accuracy and precision of high-dimensional data clustering.Therefore,in view of the nonlin-ear and high-dimensional redundancy characteristics of multivariate time series data,this paper first proposes an effective and scalable hybrid feature selection clustering algorithm based on time series cloud models on the basis of the research on traditional feature selection algorithms,cloud models,and complex time series.Then,for the extracted multi-dimensional time series data features,a method combining cloud model time similarity and multi-objective particle swarm optimization algorithm is applied for feature screening and feature optimization to obtain more high-quality features,thereby effectively improving the clustering accu-racy of the hybrid algorithm.Finally,a series of simulation experiments are conducted on high-dimensional datasets.The experi-mental results show that the hybrid feature selection algorithm can effectively solve the complex feature problems of multi-dimen-sional time series data.