一种采用相空间重构的多源数据融合方法

A Fusion Method of Multisource Data Using Phase Space Reconstruction

针对化工生产系统中状态监控变量数量庞大、冗余度高等问题,提出了一种采用相空间重构的多源数据融合方法。该方法首先根据互信息法和Cao方法分别求取相空间重构参数延迟时间和嵌入维数;然后,基于信息熵对自适应加权融合估计方法的融合目标函数进行改进,并利用社会认知优化算法确定各信息源的权重系数,实现多源数据融合;最后,通过实际化工生产系统的数据分析对所提方法进行有效性验证。实验结果表明,相比于传统方法,由该方法得到的重构相空间的信息更加完备,其信息量和平均峰值信噪比分别平均提高135.6%和40.6%。该方法为解决多源异类传感器数据融合问题提供了一种新思路。

A new fusion technology for multi-source data based on the phase space reconstruction is proposed to focus on the problem of multivariable and high redundancy of the condition monitoring variables in the chemical production system. Both the mutual information method and the Cao method are used to select the reconstruction parameters, the time delay and the embedding dimension. Then, the information entropy is employed to obtain an improved objective function in adaptive weighted fusion estimating method for multisouree data fusion, and the weighting coefficients of various information sources are calculated by means of a social cognitive optimization algorithm. The effectiveness of the proposed method is verified by an analysis of one case study of real chemical plant data sets. The results and a comparison with the traditional method show that the proposed method gets improvements in the amount of information and average PSNR, respectively. It is concluded that the proposed method improves the completeness of the information of the reconstructed phase space and provides a new approach for the multi-source data fusion of heterogeneous sensors.