基于ARIMA-CNN-LSTM模型的黄河开封段水位预测研究  

Prediction of water level of Kaifeng section of Yellow River based on ARIMA-CNN-LSTM model

作　　者：陈帅宇 赵龑骧[1] 蒋磊 CHEN Shuaiyu;ZHAO Yanxiang;JIANG Lei(College of Software,Henan University,Kaifeng 475000,China)

机构地区：[1]河南大学软件学院,河南开封475000

出　　处：《水利水电快报》2023年第1期15-22,共8页Express Water Resources & Hydropower Information

基　　金：大学生创新创业资助项目(20211025007)。

摘　　要：为准确预测水文条件复杂的黄河开封段水位变化,提出一种基于ARIMA-CNN-LSTM的多变量水位预测模型。该模型通过综合考虑水位的多重影响因素,结合卷积神经网络(CNN)和长短时记忆网络(LSTM)来学习数据中的非线性特征,同时应用ARIMA模型进行参数校正,从而实现对黄河开封段水位未来一段时间的预测。结果表明:相较于LSTM模型、CNN-LSTM模型、ARIMA模型以及BP神经网络模型,ARIMA-CNN-LSTM模型的预测精度更高,对峰值反应更灵敏,可以更加精准地预测未来一段时间的黄河开封段水位变化。Amulti-variable water level prediction model based on ARIMA-CNN-LSTM was proposed to accurately predict the water level change in Kaifeng section of the Yellow River with complicated hydrological conditions.By considering multiple factors of water level comprehensively,the model combined with convolutional neural network(CNN)and long and short time memory network(LSTM)to learn the nonlinear features in the data.Also,the ARIMA model was used for parameter correction,so as to realize the prediction of the Yellow River level in Kaifeng section for a period of time in the future.The results showed that comparing to the LSTM model,the ARIMA model and the classic BP model,the ARIMA-CNN-LSTM model has greater prediction accuracy,was more sensitive to peak response,and could more accurately anticipate the future water level changeof the Yellow River level in Kaifeng section.

关 键 词：水位预测 时间序列分析 ARIMA CNN LSTM 黄河 

分 类 号：P338.9[天文地球—水文科学] TP391.9[水利工程—水文学及水资源]