Ocean station water level space-time prediction method and device based on deep learning

Abstract

The invention discloses an ocean station water level space-time prediction method and device based on deep learning. The method comprises the steps: acquiring observation data of multiple points of water levels of an ocean station to be predicted, wherein the observation data of the multi-point water level has a space-time mapping relation; inputting observation data of multi-point water levels into the CNN and LSTM deep learning model trained in advance, wherein the CNN model is used for extracting water level spatial feature data, and the LSTM model is used for extracting water level time characteristic data corresponding to the water level space characteristic data; and based on the water level spatial feature data and the water level time feature data, outputting a water level prediction result of the to-be-predicted ocean station through a full connection layer. According to the method, high-precision forecasting of ocean station water level data can be realized; only water level sequence data of a plurality of ocean stations are needed, and other data are not needed; occupied resources are few, and the calculation speed is high. And the method can be used for but not limited to ocean station water level forecasting, and can also be used for forecasting other elements except the water level element.

Description

technical field [0001] The invention relates to the technical field of marine science and technology, in particular to a deep learning-based spatiotemporal water level prediction method and device for marine stations. Background technique [0002] Water level is one of the most important components of the marine environment. Water level prediction is of great significance in the fields of marine transportation, disaster prevention and mitigation, ecological protection and energy utilization. The most commonly used water level prediction method is the harmonic analysis method. This method uses the Fourier series to expand the measured water level data, and separates the astronomical tide part of the water level from each tidal tide, so as to realize the astronomical tide through the prediction of the tidal equinox. Forecasting, since most of the water level signals are composed of astronomical tide signals, the prediction of astronomical tides can basically be equal to the p...

AI Technical Summary

Problems solved by technology

Another prediction method is to use machine learning to learn the law of water level changes, so as to predict the water level. The BP neural network model is commonly used, but most of the existing neural network models have simple structures and single types, which have great limitations.

Images