The use of Artificial Neural Networks For Modeling of Removal of Heavy Metals From Water through Phytoremediation

Heavy metals are one of the environmental pollutants and human s contact with some of them can cause chronic and sometimes acute poisoning. Among the methods used to clean up contaminated sites, phytoremediation is considered as a cost-effective and biocompatible option. Phytoremediation refers to a group of technologies that use plants to reduce, remove, analyze and consolidate environmental toxins. The modeling of phytoremediation processes can be used as a safe tool to save the economy and avoid repeated testing. Among the methods of modeling, artificial neural network is accurate and widely used in biotechnological processes. Results of the study showed that correlation coefficient reached 1 indicating that there is a good match between actual values and those predicted by modeling. From a total of 30 data, two-thirds of data was used to train a network and the remaining third was used to assess the modeling accuracy. The middle transition function purelin, output transfer function tansig and the number of neurons (five) were determined as the best parameters to train the network. The error rate of network training was estimated to be 0.0511 and error evaluation of the network accuracy was found to be 0.763.

کلیدواژه ها:

phytoremediation ، heavy metals ، modeling ، artificial neural networks ، regression