|
چکیده
|
Investigating the impact of effective factors on membrane properties using traditional methods is expensive and time-consuming. Intelligent modeling in this field can be beneficial by eliminating experimental limitations and reducing costs. Modeling the performance of the heterogeneous cation exchange membranes in terms of ion exchange capacity (IEC) and permselectivity was investigated in the current study. So, the dataset of the IEC and permselectivity of the heterogeneous cation exchange membranes were collected over a broad range of conditions, including solvent /polymer ratio, polymer kind, and six different types of additives. Depending on the type of additive, its addition in different concentrations improves the electrochemical properties of the membrane. Three types of machine learning algorithms, i.e., radial basis functions (RBF), gaussian process regression (GPR), and multilayer perceptron (MLP) were applied to model the IEC and permselectivity. A competitive assessment of the established models for IEC and permselectivity indicated that the RBF approach overcomes the others. The RBF model for permselectivity provides excellent outcomes for the train, the test, and all data with MAPEs of 0.00%, 3.189%, and 0.697%, respectively. The RBF model for IEC shows MAPEs of 0.00%, 7.455%, and 1.631% for the train, test, and all data, respectively. The capability of the novel models for describing the IEC and permselecyivity was examined under several operating conditions, and favorable results were observed.
|