• 关键词：电-Fenton  H₂O₂  数学模型  BP神经元网络

• 基金项目：国家自然科学基金资助项目(No.50478049);广东省自然科学基金资助项目(No.04011215)

• 姜成春
• 深圳职业技术学院建筑与环境工程学院, 深圳 518055

• 张佳发
• 华侨大学土木工程学院, 泉州 362021

• 李继
• 哈尔滨工业大学深圳研究生院城市与土木工程学科部, 深圳 518055

• 摘要：以Pt为阳极,石墨碳棒为阴极,Na₂SO₄为支持电解质,实验探讨了电化学原位产生H₂O₂的规律.通过正交试验,确定阴极溶液初始pH值、电流密度CD、通氧流量Q和支持电解质浓度C_Na2SO4等主要参数对H₂O₂产生量的影响,并提出最佳参数组合pH=2.00,CD=1.02mA·cm²,Q=0.4L·min^-1,C_Na2SO4=0.1mol·L^-1,极间距D=6cm.采用二次多项式逐步回归和BP神经元网络2种方法,建立了这些参数对于H₂O₂产生量的预测模型,并对模型进行检验.结果表明,2种方法在一定参数条件下都可预测阴极区溶液中H₂O₂浓度,BP神经元网络法预测的准确度好于二次多项式逐步回归方法,且更适合于在线控制.

• Abstract：In an electrochemical reactor using Pt as anode, graphite as cathode and Na₂SO₄ as supporting electrolyte, the in situ generation of H₂O₂ electrochemically was investigated. According to the orthogonal test, the main influencing factors on the H₂O₂ generation, including pH, current density, oxygen flow and concentration of supporting electrolyte, were quantified, and the optimal combination of these factors was proposed. Then, two methods including regression analysis and BP artificial neural net works, were employed to model the H₂O₂ generation that including the influencing factors. Also the models were evaluated, and the results showed that the H₂O₂ concentration of the cathode region could be predicated under certain conditions by both of the mathematic models. Moreover, the prediction accuracy of BP artificial neural net works was higher than that of regression analysis, and the former was more feasible to apply in on-line control system.

• Key words：electro-Fenton  hydrogen peroxide  quantification model  BP artificial neural net works

• 
  摘要点击次数： 2264 全文下载次数： 3133