Advanced Kernel function-based models and soft computing techniques for optimizing oxygen transfer efficiency in solid jet aerators with circular openings: a comparative study of Gaussian process regression, random forest and support vector machine approaches

This research delves into the optimization of Oxygen Transfer Efficiency (OTE) for solid jet aerators featuring circular openings by applying a range of advanced soft computing methodologies. The study considers a diverse set of operational variables, including the number of jets (1, 2, 4, 8), jet lengths (170 mm, 270 mm, 370 mm, 470 mm), and discharge rates (1.05 l/s, 1.54 l/s, 1.93 l/s, 2.46 l/s, 3.04 l/s). By integrating empirical data with soft computing techniques such as Random Forest (RF), Gaussian Process Regression (GPR), Support Vector Machine Regression (SVMR), and Non-Linear Regression (NLR), the research significantly enhances OTE predictive accuracy. Performance parameters, including coefficient of correlation (C.C), Nash–Sutcliffe efficiency (NSE), root mean square error (RMSE), scattering index (SI), and mean absolute error (MAE), were evaluated. Results indicate that Gaussian Process regression with a radial basis kernel outperforms other techniques with 1.0000 and 0.9994 (CC), 0.0364 and 0.1242 (MAE), 0.0567 and 0.2305 (RMSE), 0.0107 and 0.0390 (SI), and 0.9999 and 0.9987 (NSE) for training and testing datasets. Sensitivity analysis shows discharge significantly influences OTE, with CC value of 0.0683, MAE of 5.025, and RMSE of 6.8018. The number of openings also impacts OTE. The findings underscore the GPR_RBF model's robustness, offering valuable insights for optimizing aeration systems in wastewater treatment and related industrial applications. This study contributes to the broader field of environmental engineering by demonstrating the application of AI-driven models to address real-world complexities in aeration processes.