文献类型：会议论文

英文题名：Ensemble Tidal Prediction Scheme by Combining Harmonic Analysis and Meteorological Predictive Module

作者：Wang, Rui[1];Yin, Jianchuan[1,2,3];Xu, Dongxing[1,2,3]

机构：[1]Guangdong Ocean Univ, Zhanjiang 524009, Peoples R China;[2]Guangdong Prov Key Lab Intelligent Equipment Sout, Zhanjiang 524088, Peoples R China;[3]Guangdong Prov Engn Res Ctr Ship Intelligence & S, Zhanjiang 524088, Peoples R China

会议论文集：5th International Conference on Neural Computing for Advanced Applications (NCAA)

会议日期：JUL 05-07, 2024

会议地点：Asia Pacific Assoc Cognit Intelligence, Guilin, PEOPLES R CHINA

主办单位：Asia Pacific Assoc Cognit Intelligence

语种：英文

外文关键词：Harmonic constant method; Ensemble prediction method; Correlation analysis; Adaptive module Adjustment; Long short-term memory network

外文摘要：Attributing to the influences of complex environmental factors on tides, there are often large errors in tidal prediction. To improve the accuracy of tide prediction in practical application, we apply the multi-module combinatorial prediction method to tidal prediction by taking consideration of the correlation between tidal level changes and air pressure, air temperature. By jointly modeling air pressure and air temperature data and tidal data, the empirical mode decomposition (EMD) method can extract features that affecting dynamic tidal changes. The harmonic constant method is firstly employed to predict the periodical changes of tides, then the air pressure and air temperature data are processed by the EMD method to extract the features that affecting the dynamic tidal changes. Then, the difference between the predicted value by harmonic method and the measured values was decomposed at multiple levels, and the long and short-term memory network (LSTM) model was used to predict the dynamic tidal changes between the measured tidal level and the periodical changes predicted by harmonic method. In the model, the effects of air pressure and air temperature on tides are considered and adjusted by the model parameters adaptively to accommodate the effects of complex environmental factors on tides. Finally, the tidal prediction results of the harmonic constant model are combined with those of the modular method. Tidal prediction simulation was conducted based on the Tidal stations of Canaveral and Old Port Tampa, the results show that the multi-module combination prediction method exhibits better performance than the traditional harmonic analysis method, providing a new approach for tidal prediction in areas of Marine safety management, Marine science research, etc.