| Along with the progress of society and the development of science and technology, in th field of petrochemical industry, pipeline network is usually very complicated, the diameter size of pipe and structure of pipe are not the same. All of this forming a huge and complex of pipeline system which is connected to each other. Like the vast and complex pipe network system whome used to dynamic simulation, the problem of to design the computing platform with real-time computation is the real-time and complexity. But the real-time is very difficult to achieve, so to achieve real-time with two aspects in this paper, one is to set the appropriate initial values, so that we can use less time to finish calculation iteration, on the other hand we instroduce the concept of agent when program designed, and to use Hadoop to computing iteration with distributed method.Firstly, it uses the method of graph theory to calculate the pipe network, according to the topology of the network to permutation transform correlation matrix which can't achieve the requirement. So in this paper uses the linear initial value method which use the superposition principle to set the initial value, and with a simple estimate can get a close value. Then after a large number of experimental verification of the method to prove the result which shorten the number of iterations and reduce the time required for the calculation.Secondly, for petrochemical field of pipe network with using graph theory to calculate pipeline network simulation. And this process may produce large matrix. So in the process of calculating iteration which involves calculation of large matrix.And there are many large and complex pipes which with many other equipment.And for simulation platform,if it is a single computer,it can not calculate the result fastly.In this paper it used the Hadoop which is mature and open source distributed computing framework for the operation of pipe network graph theory.Hadoop connection device can be used to calculate distribution of load balancing which means to allocate the large amount of calculation to the node of cluster which is free.Thirdly, the network simulation platform base on this paper that is for students or workers to learning online on training simulation platform. The architecture of distributed network simulation platform and the server is now nginx plus tomcat which becomes the architecture of distributed. Database used the Mysql and HBase, including mysql as a relational database for transaction processing, hbase is geared to the needs of columns in the database. It can store large amounts of data, designing program calculation to introduce the concept of agent, and agent based modeling and classification.Lastly, it used the the methyl acrylate (MA) in one factory as the model or example to exlpain process that is to a simplified seven rectifying column with formation of the rectifying column group and the calculation of pipe network between them as a case analysis. It makes full use of the UML software design and design patterns to case, case analysis and research which bases on the distributed network simulation platform. |
PDF Full Text Request