Objective:The aim of the present research was to model the removal of lead and cadmium from the aqueous solution by nanocomposite prepared on the basis of mineral stabilizer,using artificial neural network model and ANN optimized with particle swarm optimization algorithms.

Methods:Adsorption studies were carried out in a batch manner by considering different parameters such as the initial pH of the solution,adsorbent dosage,contact time,initial concentration of the pollutants and temperature.These parameters were considered as the input of the models and the efficiency of pollutant removal was also considered as the output of the models.

Results:The results showed that the accuracy of the ANN-PSO model was higher than the ANN model, so that the value of the R2 statistic in the ANN model for removing lead and cadmium was 0.91 and 0.90, respectively,and in the ANN-PSO model, it was 0.96 and 0.95, respectively.Examining the convergence of the PSO optimization algorithm showed that in the first iterations where the population particles are randomly selected, the optimization error is very high in all cases. Over time, with the global search and convergence to the global optimal limits, the weighted inertia increases, and the slope of the changes of the objective function in the convergence graph decreases over time, until finally optimization takes place.

Conclusions:In general, the use of ANN-PSO model can provide fast and acceptable predictions about the effect of various factors on the removal process of pollutants and is a suitable and economical alternative in terms of cost and time for laboratory studies.