Random forest model for removal of bromophenol blue using activated carbon obtained from Astragalus bisulcatus tree

详细信息    查看全文

• 作者：M. Ghaedi^a ; ^{m_ghaedi@mail.yu.ac.ir" class="auth_mail} ; A.M. Ghaedi^b ; ^{abm_ghaedi@yahoo.com" class="auth_mail} ; E. Negintaji^a ; A. Ansari^b ; A. Vafaei^b ; M. Rajabi^c
• 关键词：Bromophenol red (BPB) ; Activated carbon ; Adsorption
• 刊名：Journal of Industrial and Engineering Chemistry
• 出版年：25 July 2014
• 年：2014
• 卷：20
• 期：4
• 页码：1793-1803
• 全文大小：3557 K

文摘

In this research, activated carbon (AC) simply was prepared from a local, abundant tree in south of Iran. The AC with low cost and toxicity is a good candidate for bromophenol blue (BPB) removal from aqueous media. The AC with nano scale pore diameter is applicable for this dye removal following optimization of the influence of various parameters including contact time, pH, initial dye concentration and amount of adsorbent. Subsequently, experimental data was analyzed by four kinetic models including pseudo first and second-order, Elovich and the intraparticle diffusion equations and subsequently their respective parameters such as rate constants, equilibrium adsorption capacities and correlation coefficients was investigated and based on well known criterion their applicability was judged. The result shows that adsorption of BPB onto proposed adsorbent at all conditions such as versatile adsorbent dosages and initial BPB concentrations sufficiently described by the combination of the pseudo second-order equation and interparticle diffusion model. It was found that equilibrium rate of the BPB adsorption at various adsorbent dosage well fitted by Langmuir. Investigation of experimental result by two approaches (multiple linear regressions (MLR) and random forest (RF)) models show that RF is a powerful tool for prediction of BPB adsorption by activated carbon obtained from Astragalus bisulcatus tree. The optimal tuning parameters for RF model are obtained based on the n_tree = 100, m_try = 2. For the training data set, the MSE values of 0.0006 and the coefficient of determination (R²) values of 0.9895 for RF model and the MSE value of 0.0104 and the R² value of 0.823 for MLR model are obtained.