Research On Key Technology Of Data Transmission Based On Data Middle Platform

With the promotion of digital transformation,the electric power industry is also exploring how to use big data,artificial intelligence and other technologies to improve efficiency and service quality,and the data middle platform has become an important trend in digital transformation.Although electric power enterprises are accelerating the construction of data middle platform in recent years,there are still some problems in data transmission,such as rough data classification,inability to effectively transmit data according to service requirements,and data transmission efficiency.cannot fulfill the requirements of real-time.Therefore,this dissertation designs and builds a data transfer model based on data middle platform,which improves the data transfer efficiency while meeting the changing business requirements by combining dynamic priority-based data division rules with deep reinforcement learning technology.A data transfer display system is also implemented,using visualization techniques to show the changes of indicators during data transfer more intuitively,which facilitates researchers to compare and optimize the data transfer model and improve the data transfer capability of the data middle platform.In this dissertation,the following work is mainly carried out:Firstly,for the characteristics of different types of data in power system,the importance of data is assessed from three dimensions of data characteristics,business requirements and waiting time,and a data classification rule based on dynamic priority is designed.The division rule fully considers the characteristics of data urgency,real-time,reliability,security,and the category to which they belong,and ensures timely transmission of relatively important data and avoids starvation as much as possible while satisfying the business needs.Secondly,in order to improve the efficiency of data transmission,shorten the transmission time and enhance the load balancing capability,this dissertation constructs a data transmission model using the SAC-Discrete scheduling algorithm.To verify the performance advantages of this model,it is experimentally compared with FCFS and DDPG scheduling algorithms,and the results prove that the model has certain superiority in three indexes:makespan,average transmission completion time and load balancing capability.Thirdly,in order to observe the data transmission process more intuitively and compare the performance differences of different models so as to further optimize the data transmission model,a data transmission display system is designed and implemented based on the above research results.The system aims to present key information in the data transfer process through visualization so that researchers can accurately analyze the advantages and disadvantages of various data transfer models.