Data Management And Implementations For Distributed Electromechanical Systems Based On Consistent Hashing Algorithms

The ability to efficiently and reliably store and manage the operational data of distributed electromechanical systems is a prerequisite and guarantee for system health monitoring and operation and maintenance.With the rapid development of electromechanical equipment for intelligent integration,high precision of IOT and remote measurement and control,the number of edge devices with computing and storage capabilities is increasing,making it difficult for the existing distributed storage model to efficiently manage the huge amount of data generated by edge nodes,and the efficient management of data load balance has become a key issue to be solved.Based on the analysis of the actual requirements for data management of distributed electromechanical systems,the thesis effectively reduces data redundancy through data pre-processing techniques and proposes a distributed data storage method based on improved consistent hashing algorithms to achieve efficient data storage,and tests the effectiveness of the proposed method in combination with the designed data management platform for distributed electromechanical systems.The main research contents of the thesis are as follows:The analysis focuses on the functional requirements of the data management system from two aspects of data pre-processing and storage,clarifies the signal characteristics and data structure of the distributed electromechanical system,and ensures real-time data acquisition and transmission through embedded FPGA data acquisition technology,based on which the overall framework of the data management platform of the distributed electromechanical system is designed.In order to reduce the pressure of data transmission and data storage load on the server terminal,the data is pre-processed at the edge node through data integration and data cleaning,in which the standardised integration of heterogeneous data from multiple sources is achieved by using data coding techniques.The data is cleaned by using a variational modal decomposition method to extract the real operating signal and the noise signal.At the same time,the System Link architecture enables efficient remote transmission of data.Aiming at the unbalanced storage load caused by the massive data of distributed electromechanical systems,a data storage method based on the proportional distribution of weights is proposed in combination with the consistent hashing algorithm.Firstly,the mapping relationship between host nodes,operational data and hash rings is established based on the consistent hashing algorithm,and the addressing rules for data storage are clarified.Secondly,considering the differences in performance of each host,virtual nodes are allocated based on weights,and on the basis of solving the hash offset problem,load adaption is achieved,so as to establish a dynamic allocation mechanism of virtual nodes.Finally,the data storage function is improved by combining with HDFS database.The data management platform of the distributed electromechanical system was tested in an experimental environment.The test results show that the designed data management platform can effectively collect and pre-process the operational data of the distributed electromechanical system,and the resulting reconstructed data can improve the signal-to-noise ratio by about 21 d B and reduce the total harmonic distortion rate by about 3.8% compared with the original data.The proposed data storage method shows excellent load balancing compared to the traditional consistent hashing algorithm,which can effectively reduce the number of dynamic allocations and enable efficient data storage.