Research On Big Data Analysis Of Marine Environment Based On Long Short-term Memory Network And Regional Ocean Model

Ocean observation is affected by time and space,and the acquired ocean data is sparse in space and discontinuous in time.The ocean numerical model simulates ocean motion by solving the equations of physics and mathematics,and is widely used in ocean environment simulation and numerical prediction.Because the physical motion and interaction in the ocean are complex,the ocean numerical model cannot fully describe the physical phenomena and laws of motion,so there are certain errors in the prediction of the ocean numerical model.This thesis uses machine learning to correct the errors of ocean numerical models,uses big data technology to store and analyze ocean data,and realizes the effective use of data.The main work of this thesis is as follows:1)This thesis proposes an error correction model based on long and short-term memory neural network to reduce the error between the forecast value of the ocean numerical model and the actual observation value,and to improve the forecast accuracy of the ocean numerical model.Analyze the causes of errors in the numerical model,and train the model through historical observation data and the predicted value of the numerical model.Experiments show that the model can effectively correct the prediction errors of ocean numerical models.2)Aiming at the problem of large amount of marine data and difficult to store,this thesis proposes to build a marine environment big data storage system based on HBase,and uses big data technology to mine the value information in marine big data.Give full play to the read and write advantages of distributed storage,and optimize regional data query.A parallel association rule mining algorithm is proposed to solve the problem of real-time mining of ocean data association rules.Experiments show that the storage system proposed in this thesis has the characteristics of high throughput and low latency,and is suitable for storing big data in the marine environment;the parallel association rule mining algorithm proposed can effectively mine the association rules of the marine environment.