Application of hybrid artificial neural network (ANN)–particle swarm optimization (PSO) for modelling and optimization of the adsorptive removal of cyanide and phenol from wastewater using agro-waste-derived adsorbent

Abstract In the present study, the waste part of the banana tree was used as a precursor, and copper chloride salt was used as an impregnating agent for the preparation of adsorbent to remove both cyanide and phenol from synthetic wastewater. Initially, thermogravimetric analysis was used to determine the rate of carbonization of the material with temperature, and thus, the optimum temperature (370 °C) and time of carbonization (35 min) were assessed. Different samples of adsorbents were prepared next by varying the weight ratio of pseudo-stem of waste banana tree to copper salt from 1:1 to 30:1. All the samples were then tested for removal of both the pollutants, and the ratio (20:1) corresponding to maximum removal of both the pollutants was considered as optimum. Therefore, further studies were conducted with the adsorbent prepared at optimum ratio, temperature and time and such adsorbent was termed as copper impregnated activated banana tree (CIABT). One variable at a time approach was followed to find out the most effective condition based on the maximum removal of pollutants. Maximum removal of 95.99 ± 1.03% and 97.33 ± 0.04% was achieved for cyanide (initial concentration: 100 ppm) and phenol (initial concentration: 450 ppm), respectively, at an optimum contact time of 150 min, the particle size of 90 μ, the adsorbent dosage of 10 g/L, pH 8.0 using CIABT at 25 °C. Hybrid artificial neural network–particle swarm optimization were employed for modelling-optimization of removal of both the pollutants while achieving 91.4–99.99% and 86.43–99.99% removal of cyanide and phenol, respectively, from simulated wastewater.