基于TCN-BiLSTM与LSTM模型对比预测北洛河径流

Comparative prediction of runoff in the Beiluo River,Shaanxi Province of northwestern China based on TCN-BiLSTM and LSTM models

【目的】本研究旨在探究TCN-BiLSTM耦合模型与传统LSTM模型在径流模拟预测中的性能,为洪水风险管理和区域水资源规划提供准确有效的径流预测模型。【方法】以北洛河流域为研究区,基于双向长短期记忆网络(BiLSTM)和时域卷积网络(TCN)建立一种新的径流预测耦合模型TCN-BiLSTM。利用相关性分析,筛选预测径流的输入因子,确定4种不同的输入方案应用于TCN-BiLSTM耦合模型和传统LSTM模型,每个模型分别预测1、2、3 d的径流量。采用平均绝对误差(MAE)、均方根误差(RMSE)和拟合优度(R^(2))来评估模型的预测性能。【结果】(1)TCN-BiLSTM耦合模型整体预测性能优于LSTM模型,TCN-BiLSTM模型R^(2)达到0.91,高于LSTM的0.89。相比于LSTM,TCN-BiLSTM对于峰值和突变点的捕捉能力更强,对于波动大的复杂数据预测效果更优;(2)在针对未来1~3 d径流量预测中,随着预见期的延长,4种方案下TCN-BiLSTM和LSTM模型的预测效果均有所下降,相较于预测1 d,预测3 d的TCNBiLSTM和LSTM模型的R^(2)分别平均下降了0.17和0.14,RMSE分别平均增大了4.59和4.40,MAE分别平均增大了1.26和1.31;(3)在4种输入方案里,日累积降水量和日径流量作为输入变量时,模型的预测效果最好。降水数据的加入使得TCN-BiLSTM和LSTM模型相较于单一日径流数据作为输入变量时,1、2、3 d径流量预测的R^(2)分别提高15%、14%、6%和18%、14%和1%。【结论】TCN-BiLSTM耦合模型和LSTM模型R^(2)均能达到0.85以上,TCN-BiLSTM模型R^(2)较LSTM提高了2%。对比来看,TCN-BiLSTM模型在拟合洪水过程中表现更为优异,对于汛期的预测性能优于非汛期。输入变量对模型的影响较大,有效且高质量的气象数据能够提高模型的预测性能。

[Objective]The aim of this study was to investigate the performance of coupled TCN-BiLSTM model and traditional LSTM model in runoff simulation and prediction,specifically the effect of different input variables on the accuracy of machine learning hydrological model and performance of the model under different foresight periods.[Method]A new coupled model TCN-BiLSTM for runoff prediction was established based on bi-directional long short-term memory network(BiLSTM)and temporal convolutional network(TCN)with the Beiluo River Basin as the study area.Using correlation analysis,the input factors for predicting runoff were screened,and four different input schemes were identified to be applied to the coupled TCN-BiLSTM model and the conventional LSTM model,each of which predicted runoff volumes for 1,2,and 3 d,respectively.Mean absolute error(MAE),root mean square error(RMSE)and goodness of fit(R^(2))were used to assess the predictive performance of the model.[Result](1)The overall prediction performance of the TCN-BiLSTM coupled model was better than that of the LSTM model,and the R^(2) of TCN-BiLSTM can reach 0.91,which was higher than that of the LSTM,0.89.Compared with LSTM,TCNBiLSTM was more capable of capturing peaks and mutation points,and was better at predicting complex data with large fluctuations.(2)In the runoff prediction for the next 1−3 d,the predictive effectiveness of TCN-BiLSTM and LSTM models under the four scenarios decreased with the extension of the foresight period,and compared with the prediction of 1 d,the TCN-BiLSTM and LSTM R^(2) for the prediction of 3 d decreased on average by 0.17 and 0.14,respectively,and the RMSE increased on average by 4.59 and 4.40,respectively,and the MAE increased on average by 1.26 and 1.31,respectively.(3)Among the four input scenarios,the best model predictions were obtained when daily precipitation data and daily runoff data were used as input variables.The inclusion of precipitation data improved the R^(2) of the TCN-BiLSTM and LSTM models for 1,2,and 3 d runoff predictions by 15%,14%,6%,and 18%,14%,and 1%,respectively,compared with the single daily runoff data as an input variable.[Conclusion]Both the TCN-BiLSTM coupled model and the LSTM model R^(2) can reach more than 0.85,and the TCN-BiLSTM R^(2) is improved by 2%compared with the LSTM.In comparison,the TCN-BiLSTM model performs better in fitting the flood process,and the prediction performance for flood season is better than that for non-flood season.The input variables have a greater impact on the model,and effective and high-quality meteorological data can improve the prediction performance of the model.