Genetic algorithm hyper-parameter optimization using Taguchi design for groundwater pollution source identification

Abstract

Groundwater pollution has been a major concern for human beings, since it is inherently related to people's health and fitness and the ecological environment. To improve the identification of groundwater pollution, many optimization approaches have been developed. Among them, the genetic algorithm (GA) is widely used with its performance depending on the hyper-parameters. In this study, a simulation–optimization approach, i.e., a transport simulation model with a genetic optimization algorithm, was utilized to determine the pollutant source fluxes. We proposed a robust method for tuning the hyper-parameters based on Taguchi experimental design to optimize the performance of the GA. The effectiveness of the method was tested on an irregular geometry and heterogeneous porous media considering steady-state flow and transient transport conditions. Compared with traditional GA with default hyper-parameters, our proposed hyper-parameter tuning method is able to provide appropriate parameters for running the GA, and can more efficiently identify groundwater pollution.

Predict initial subthalamic nucleus stimulation outcome in Parkinson's disease with brain morphology

“… 25 The randomly selected hyperparameters used in the SVM could lower the accuracy of prediction; in other words, the performance of the machine learning algorithm could be largely affected by these randomized hyperparameters. Therefore, a genetic algorithm approach, an evolutionary computation algorithm, 26 was adopted in the present study to obtain optimized hyperparameters (cost and gamma) in machine learning. Finally, the model (radial basis function [RBF] kernel) established with these features and hyperparameters (after scaled) was further used for predicting the test set.…”

Genetic Algorithm in Multimedia Dynamic Prediction of Groundwater in Open-Pit Mine

“…Mohammadrezapour and Kisi used neural network models to select precipitation, evaporation, surface runoff, and other information related to shallow groundwater dynamics and used them to predict phreatic groundwater dynamics [ 8 ]. Xia et al successfully predicted the groundwater dynamics of semiconfined aquifers using the neural network method based on extraction volume and hydrometeorological factors [ 9 ]. On the basis of the current research, through the design and calculation process of MATLAB 7 platform, taking the monitoring wells distributed in an open-pit mining area as an example, the short-term prediction of groundwater dynamics in the study area is carried out by using BP neural network model and BP neural network model based on genetic algorithm.…”

Groundwater contamination source identification based on Sobol Sequences-based Sparrow Search Algorithm with a BiLSTM surrogate model

“…So far, the most widely used optimization algorithms are gradient-based traditional optimization algorithms and heuristic optimization algorithms. Heuristic optimization algorithms, such as simulated annealing algorithms (Jha and Datta, 2013), genetic algorithms (Xia et al, 2019), and particle swarm optimization algorithms (Guneshwor et al, 2018), are faster and more effective than traditional optimization algorithms. The sparrow search algorithm(SSA) was first proposed by Xue and Shen(2020), which is inspired by the foraging and antipredatory behaviors of sparrows and has high convergence performance and local search capability.…”