Design And Research Of Marine Ranching Big Data Platform Based On Hadoop Architecture

Marine ranching is an important part of China’s fishery,and its healthy development is conducive to the restoration of Marine ecological environment,recovery of Marine fishery resources and maintenance of sustainable development of Marine fishery.With the development of modern information technology,all kinds of data produced by Marine ranches are growing at a rapid rate that can be seen by the naked eye.As a result,the data cannot be stored in a timely and effective manner,resulting in incomplete data storage and failure to discover the relationship between the data in a timely manner.Not fully exploiting the potential value of stored data.In recent years,due to the rapid development of big data,effective methods have been provided to solve the problems of data storage "fragmentation",low data value utilization,and data "islands".In response to the above problems,this paper based on the Hadoop distributed infrastructure,researched and designed a big data platform for marine ranches,which effectively improved the ability and information level of marine ranching data management.First,it introduces the development status and existing problems of marine ranches and big data at home and abroad,and then elaborates the key technologies related to the construction of marine ranches big data platform.These key technologies include the Hadoop ecosystem,artificial neural networks,and fruit fly optimization algorithm principles and detailed improvement processes,which lay the foundation for the design and research of building a big data platform for marine ranches.Then,by analyzing the data sources,actual needs and big data standardization system of the marine ranching,the overall architecture,standard system and overall framework of the marine ranching big data platform were designed and realized.This paper proposes the use of improved fruit fly optimization algorithm(IFOA)to optimize the BP neural network algorithm and establishes a marine pasture dissolved oxygen prediction analysis model(IFOA-BP),in order to study the nonlinear relationship between marine pasture water quality factors.We verified the prediction accuracy of the model through the analysis of the data from Zhangjialou Marine Ranch in Shandong Province.Using the MapReduce research in the Hadoop distributed architecture,the detailed parallel calculation process of the BP neural network was designed to realize the parallel operation of the BP neural network and solve the analysis and processing of the massive data of the marine ranching.We analyzed the classification accuracy and efficiency of the neural network parallelization algorithm through an actual marine pasture water quality data example.Finally,the goals,business functions,databases,and data processing procedures of the marine ranch big data platform are designed according to the characteristics of the marine ranch data and actual needs.The marine ranch big data platform that realizes marine ranch data collection,data storage management,data analysis and calculation,and data application solves the problems of "fragmentation" of marine ranch data,low data value utilization,and data "islands".