Ocean salinity intelligent prediction model based on particle swarm optimization LSTM neural network

Abstract

Propose a prediction model of salinity time series model based on PSO particle swarm optimization for LSTM neural network. In order to solve the problem that some hyperparameters are difficult to determine and the efficiency is low in the classic LSTM neural network, this paper introducing PSO algorithm model into small-scale training of preset neural network and hierarchical structure of LSTM neural network training is determined and given in the iteration, the experimental results are compared with those of the original LSTM neural network model in addition. The model is used to predict and verify the actual salinity time series in the sea area near a certain coordinate and shows the feasibility and optimization.