Multi-cluster high performance computing method based on multimodal tensor in enterprise resource planning system

The big data representation and processing method based on multimodal tensors can achieve the fusion representation of different types of data, and perform correlation analysis on the fusion data through multimodal tensor decomposition, providing accurate data services for practical applications. However, in the process of big data analysis and processing based on multimodal tensors, there are still some urgent problems to be solved, such as repeated calculations caused by streaming data, time-consuming calculations caused by large scale, ultra-high energy consumption caused by complex calculations, and poor analysis results caused by complex relationships. This article uses big data tools Hadoop and Spark to build a data mining platform to analyze and study the data of traditional manufacturing enterprises' enterprise resource planning (ERP) systems. In order to improve the accuracy of data analysis, the Stacking fusion framework was also introduced. Furthermore, We transform the data generated by the enterprise into valuable information, making more accurate decisions for enterprise personnel. This article proposes a distributed parallel multi-mode tensor chain decomposition method. In the distributed parallel multi-mode tensor chain decomposition method, a method and theoretical proof for solving tensor chain decomposition are proposed based on the original tensor chain decomposition results and tensor chain decomposition results. The multi kernel clustering algorithm based on K-Means aims to improve the performance of kernel K-clustering by integrating a set of kernels. Deep learning model is also proposed to fuse these multilinear attributes, a clustering method based on weighted average similarity matrix and a clustering method based on optional weighted Euclidean distance were proposed. The distributed parallel multimodal method proposed in this article can effectively achieve tensor chain decomposition, thereby improving the efficiency of big data processing. © 2001 Elsevier Science. All rights reserved