基于EMD-LSTM的水产养殖水温预测模型  

Aquaculture Water Temperature Prediction Model Based on EMD-LSTM

作　　者：张锦荣 王康谊[1] ZHANG Jinrong;WANG Kangyi(School of Information and Communication Engineering,North University of China,Taiyuan 030051)

机构地区：[1]中北大学信息与通信工程学院,太原030051

出　　处：《计算机与数字工程》2023年第9期2189-2194,共6页Computer & Digital Engineering

摘　　要：针对水温预测问题,为提高水产养殖过程中水温的预测精度,提出了一种基于EMD-LSTM的水温预测模型。首先,利用Pearson相关系数检验法分析各水质数据与水温之间的相关度,确定预测模型的输入变量;采用EMD算法对原始水温数据进行多尺度分解,对各分量进行LSTM训练建模,最后叠加求和各分量预测值,实现水温序列的预测。利用海南省某水产养殖基地采集到的5项指标(水温、PH、溶解氧、盐度、空气温度)对该模型进行训练测试,并与BP、LSTM、EMD-BP三种预测模型的预测结果进行比较。结果表明,所提出的预测模型在各项评价指标上均优于上述三种模型,具有较高的预测精度,可以满足水产养殖水温精确预测的需求,为养殖水温预测预警提高辅助决策。Aiming at the problem of water temperature prediction,a water temperature prediction model based on EMD-LSTM is proposed to improve the accuracy of water temperature prediction in aquaculture.Firstly,the correlation between water quality data and water temperature is analyzed by Pearson correlation coefficient test method,and the input variables of prediction model are determined.EMD algorithm is used to decompose the original water temperature data,and LSTM training modeling is carried out for each component.Finally,the predicted values of each component are superimposed and summed to realize the prediction of water temperature series.Collected from an aquaculture base in Hainan Province five parameters(water temperature,PH,dissolved oxygen,salinity and air temperature)are used to test the model,and the results are compared with those of BP,LSTM and EMD-BP.The results show that the proposed prediction model is superior to the above three models in all evaluation indexes,and has high prediction accuracy,which can meet the demand of accurate prediction of aquaculture water temperature and improve the auxiliary decision for forecasting and early warning of aquaculture water temperature.

关 键 词：水温预测 经验模态分解 LSTM神经网络 水产养殖 

分 类 号：P332.6[天文地球—水文科学]