作者: Z. Shahryari , A. Sharifi , A. Mohebbi
DOI: 10.1134/S181023281304005X
关键词:
摘要: In this study, a three-layer feed-forward back propagation network with Levenberg-Marquardt (LM) learning algorithm was applied to predict adsorption of phenol onto activated carbon (AC). Batch experiments were carried out obtain experimental data. The neural trained considering the amount adsorbent, initial concentration phenol, temperature, contact time and pH as input parameters final desired parameter. Different transfer functions for hidden output layers different number neurons in layer tested optimize structure. An empirical equation developed by using weights optimized network. Accuracy ANN model also measured statistical parameters, such mean absolute error (MAE), square (MSE), root (RMSE) correlation coefficient (R2). Results showed that MAE, MSE, RMSE, R2 values 0.1540, 0.0565, 0.2378, 0.9998, respectively, which indicate high accuracy model. equilibrium predicted results compared data other conventional isotherm models.