eduzhai > Applied Sciences > Engineering >

Comparison of Machine Learning Methods for Predicting Karst Spring Discharge in North China

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 28 pages

Abstract: The quantitative analyses of karst spring discharge typically rely onphysical-based models, which are inherently uncertain. To improve theunderstanding of the mechanism of spring discharge fluctuation and therelationship between precipitation and spring discharge, three machine learningmethods were developed to reduce the predictive errors of physical-basedgroundwater models, simulate the discharge of Longzici Spring s karst area, andpredict changes in the spring on the basis of long time series precipitationmonitoring and spring water flow data from 1987 to 2018. The three machinelearning methods included two artificial neural networks (ANNs), namely,multilayer perceptron (MLP) and long short-term memory-recurrent neural network(LSTM-RNN), and support vector regression (SVR). A normalization method wasintroduced for data preprocessing to make the three methods robust andcomputationally efficient. To compare and evaluate the capability of the threemachine learning methods, the mean squared error (MSE), mean absolute error(MAE), and root-mean-square error (RMSE) were selected as the performancemetrics for these methods. Simulations showed that MLP reduced MSE, MAE, andRMSE to 0.0010, 0.0254, and 0.0318, respectively. Meanwhile, LSTM-RNN reducedMSE to 0.0010, MAE to 0.0272, and RMSE to 0.0329. Moreover, the decrease inMSE, MAE, and RMSE were 0.0910, 0.1852, and 0.3017, respectively, for SVR.Results indicated that MLP performed slightly better than LSTM-RNN, and MLP andLSTM-RNN performed considerably better than SVR. Furthermore, ANNs weredemonstrated to be prior machine learning methods for simulating and predictingkarst spring discharge.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×