Prediction of Methylene Blue and Congo Red Removal in Lemna minor Phytoremediation Using Neural-Wiener Modelling

Methylene blue and Congo red are synthetic dyes commonly found in industrial wastewater. These dyes can affect water quality and pose risks to human health and the environment. Phytoremediation using Lemna minor offers a simple and environmentally friendly method for reducing dye concentrations in wastewater. However, the phytoremediation process usually requires a long treatment time. Therefore, this study applied neural network-based model, namely Neural–Wiener model, to predict the percentage removal of methylene blue and Congo red during phytoremediation using Lemna minor. The model was trained using different data-splitting ratios for training, validation, and testing, as well as different numbers of hidden nodes. The results showed that the Neural–Wiener model provided a good prediction performance. The optimum model was obtained using 35% training data, 30% validation data, and 35% testing data with 140 hidden nodes. The optimum model achieved MSE, MAE, and WMAPE values of 5 × 10⁻⁴, 1 × 10⁻⁴, and 1 × 10⁻⁵, respectively, based on the percentage removal response used as the model output. These results indicate that the Neural–Wiener model is more suitable than other models for predicting the removal performance of methylene blue and Congo red in the Lemna minor phytoremediation process.