Kharazmi university , t.azari@khu.ac.ir
Abstract: (60 Views)
Accurate estimation of hydrodynamic parameters is the first step towards sustainable aquifer development. Since Theis (1935), the Type Curve Matching Technique (TCMT) has been used to estimate aquifer parameters. This method is associated with graphical errors. In this study, a supervised AI committee machine was used to eliminate errors and accurately estimate the hydrodynamic parameters of confined aquifers with high ability to approximate functions as an alternative to the conventional TCMT method and existing AI methods. In this study, pumping test data were considered as input components and the coordinates of the optimal point were considered as the output. To reduce the dimensions of the input components, the principal component analysis (PCA) technique was used. Then, the matching point coordinates were combined with the analytical solution of Theis (1935) and the values of the aquifer parameters were calculated. To develop this machine, in the first step, three ANNs with different training algorithms Levenberg–Marquardt (LM), gradient descent (GD), resilient back-propagation (RP) were developed to determine the match point and estimate the hydrodynamic parameters of the confined aquifer. Based on the modeling results, all models showed a good approximation of the hydrodynamic parameters of the confined aquifer. Then, in the second step, considering the complexity of hydrogeological systems, a committee machine consisting of three artificial intelligence models was designed and built, which used the capabilities of all three models to determine the hydrodynamic parameters of the confined aquifers. The output of the models used was combined with a supervised nonlinear combiner, and the final output of this machine was determined with very high accuracy. The results showed that the proposed committee machine model is more accurate, and better alternative to TCMT methods and artificial intelligence methods in determining the optimal match point and estimating the hydrodynamic parameters of the confined aquifer.
Keywords: Artificial neural network, supervised artificial intelligence committee machine, pumping test, type curve matching method, confined aquifer parameters.
Type of Study: Original Research |
Subject:
Hydrogeology
Received: 2025/04/15 | Accepted: 2025/05/20
Received: 2025/04/15 | Accepted: 2025/05/20
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
