UASB处理高硫酸盐废水及人工神经网络模型的构建

doi:10.19965/j.cnki.iwt.2020-1140

摘要/Abstract

摘要：

采用UASB处理高硫酸盐废水，对不同碳硫比〔m（COD）/m（SO₄^2-）=1、2、3、4.6〕条件下反应体系的处理效能进行评估，并利用神经网络模型分析不同因素对COD去除率和SO₄^2-去除率的影响。结果表明SO₄^2-去除率与m（COD）/m（SO₄^2-）成正比，而COD去除率则与m（COD）/m（SO₄^2-）成反比。当m（COD）/m（SO₄^2-）为4.6时，SO₄^2-平均去除率达98.1%，此时，COD平均去除率仅为32.2%。神经网络模型的影响因素权重分析表明进水pH、COD和m（COD）/m（SO₄^2-）为影响COD去除的主要因素，进水COD、SO₄^2-浓度和m（COD）/m（SO₄^2-）为影响SO₄^2-去除的主要因素。

关键词: 高硫酸盐废水, 厌氧处理, 神经网络

Abstract:

Sulfate-rich wastewater was treated by UASB. The performance of the reaction system under the condition of different COD/sulfate ratios[m(COD)/m(SO₄^2-)=1, 2, 3, 4.6] was studied, and the neural network model was used to analyze the influence of different factors on COD and SO₄^2- removal efficiency. The results showed that the removal rate of SO₄^2- was proportional to the m(COD)/m(SO₄^2-) and the removal rate of COD was inversely proportional to the m(COD)/m(SO₄^2-). When m(COD)/m(SO₄^2-) was 4.6, the average removal rate of SO₄^2- could reach up to 98.1%, but that of COD was only 32.2%. The weight analysis of influencing factors of neural network model shows that the influent pH, COD and m(COD)/m(SO₄^2-) were the main parameters that affect COD removal, and the influent COD, SO₄^2- concentration and m(COD)/m(SO₄^2-) were the main parameters that affect SO₄^2- removal.

Key words: sulfate-rich wastewater, anaerobic biological treatment, neural network

中图分类号:

X703

王庆宏,余静诗,梁家豪,陈春茂. UASB处理高硫酸盐废水及人工神经网络模型的构建[J]. 工业水处理, 2021, 41(7): 77-81.

Qinghong Wang,Jingshi Yu,Jiahao Liang,Chunmao Chen. Treatment of sulfate-rich wastewater by UASB and construction of artificial neural network model[J]. Industrial Water Treatment, 2021, 41(7): 77-81.

https://www.iwt.cn/CN/Y2021/V41/I7/77

图/表 6

图1 UASB实验装置图

Fig.1 Structure diagram of UASB

表1 不同阶段模拟废水组成

Table 1 Synthetic wastewater components in different periods

参数	阶段Ⅰ		阶段Ⅱ	阶段Ⅲ	阶段Ⅳ	阶段Ⅴ
运行时间	第1天—第10天	第11天—第20天	第21天—第29天	第30天—第45天	第46天—第51天	第52天—第61天
COD/（mg·L^-1）	1 000±72	2 300±131	1 000±69	1 500±75	2 300±121	2 000±116
SO₄^2-/（mg·L^-1）	1 000±32	500±28	500±28	500±16	500±12	1 000±34
m（COD）/m（SO₄^2-）	1	4.6	2	3	4.6	2

图2 UASB反应器运行性能

Fig.2 Operating performance of UASB

图3 人工神经网络模型和不同隐藏层节点数的均方误差

Fig.3 Artificial neural network model and the mean square error of the number of nodes in different hidden layers

图4 去除率实验数据与模型预测的相关性

Fig.4 Correlations between experimental datas and model pre-dictions of COD removal efficiency and SO₄^2- removal efficiency

表2 输入变量对COD和SO₄^2-去除率的相对重要性

Table 2 The relative importance of input variables for the removal efficiency of COD and SO₄^2-

输入变量	相对重要性/%
输入变量	COD去除率	SO₄^2-去除率
进水COD	22.0	22.2
进水SO₄^2-浓度	19.3	22.1
进水pH	26.3	14.3
进水m（COD）/m（SO₄^2-）	23.9	25.4
HRT	8.5	16.0

参考文献 16

1	Huang Cong , Liu Qian , Chen Chuan , et al. Elemental sulfur recovery and spatial distribution of functional bacteria and expressed genes under different carbon/nitrate/sulfide loadings in up-flow anaerobic sludge blanket reactors[J]. Journal of Hazardous Materials, 2017, 324, 48- 53. doi: 10.1016/j.jhazmat.2016.03.024
2	Choi E , Rim J M . Competition and inhibition of sulfate reducers and methane producers in anaerobic treatment[J]. Water Science and Technology, 1991, 23, 7- 9. URL
3	Katsuji U , Keiko K , Kenji I , et al. Potential availability of anaerobic treatment with digester slurry of animal waste for the reclamation of acid mine water containing sulfate and heavy metals[J]. Journal of Fermentation Technology, 1988, 66 (1): 43- 50. doi: 10.1016/0385-6380(88)90128-8
4	Lu Xueqin , Zhen Guangyin , Ni Jialing , et al. Effect of influent COD/SO₄^2- ratios on biodegradation behaviors of starch wastewater in an up-flow anaerobic sludge blanket(UASB) reactor[J]. Bioresource Technology, 2016, 214, 175- 183. doi: 10.1016/j.biortech.2016.04.100
5	Liang Jiahao , Wang Qinghong , Li Q X , et al. Aerobic sludge granulation in shale gas flowback water treatment: Assessment of the bacterial community dynamics and modeling of bioreactor performance using artificial neural network[J]. Bioresource Technology, 2020, 313, 123687. doi: 10.1016/j.biortech.2020.123687
6	Nair V V , Dhar H , Kumar S , et al. Artificial neural network based modeling to evaluate methane yield from biogas in a laboratory-scale anaerobic bioreactor[J]. Bioresource Technology, 2016, 217, 90- 99. doi: 10.1016/j.biortech.2016.03.046
7	Basu B , Singh M P , Kapur G S , et al. Prediction of biodegradability of mineral base oils from chemical composition using artificial neural networks[J]. Tribology International, 1998, 31 (4): 159- 168. doi: 10.1016/S0301-679X(97)00078-9
8	国家环境保护总局. 水和废水监测分析方法[M]. 4版北京: 中国环境科学出版社, 2005: 31- 57.
9	Zhou Jiemin , Xing Jianmin , et al. Effect of electron donors on the performance of haloalkaliphilic sulfate-reducing bioreactors for flue gas treatment and microbial degradation patterns related to sulfate reduction of different electron donors[J]. Biochemical Engineering Journal, 2015, 96, 14- 22. doi: 10.1016/j.bej.2014.12.015
10	Sabumon P C . Development of enhanced sulphidogenesis process for the treatment of wastewater having low COD/SO₄^2- ratio[J]. Journal of Hazardous Materials, 2008, 159 (2/3): 616- 625.
11	Hu Yong , Jing Zhaoqian , Sudo Y , et al. Effect of influent COD/SO₄^2- ratios on UASB treatment of a synthetic sulfate-containing waste-water[J]. Chemosphere, 2015, 130, 24- 33. doi: 10.1016/j.chemosphere.2015.02.019
12	Cheng Jun , Wang Xin , Si Tingting , et al. Ignition temperature and activation energy of power coal blends predicted with back-propagation neural network models[J]. Fuel, 2016, 173, 230- 238. doi: 10.1016/j.fuel.2016.01.043
13	Philip A , Jian Z L , Portia O B , et al. Estimation of biogas and methane yields in an UASB treating potato starch processing wastewater with backpropagation artificial neural network[J]. Bioresource Technology, 2017, 228, 106- 115. doi: 10.1016/j.biortech.2016.12.045
14	Arindam S , Divya B , Kannan P . Process integration and artificial neural network modeling of biological sulfate reduction using a carbon monoxide fed gas lift bioreactor[J]. Chemical Engineering Journal, 2020, 391, 123518. doi: 10.1016/j.cej.2019.123518
15	王爱杰, 任南琪, 甄卫东, 等. 硫酸盐还原过程中SRB的限制性生态因子的BP神经网络建模与仿真[J]. 环境科学学报, 2001, 21 (3): 267- 270. doi: 10.3321/j.issn:0253-2468.2001.03.003
16	Reyes-Alvarado L C , Okpalanze N N , Kankanala D , et al. Forecasting the effect of feast and famine conditions on biological sulphate reduction in an anaerobic inverse fluidized bed reactor using artificial neural networks[J]. Process Biochemistry, 2017, 55, 146- 161. doi: 10.1016/j.procbio.2017.01.021

[1]	顾先涛, 樊培培, 陈晓春, 高宇祥, 周仲康, 张俊杰, 马晓薇, 计巧珍, 吴妍, 徐亚艳, 董浩声, 段忠辛, 杨林, 高忠辉. 特高压换流站调相机外冷水系统腐蚀结垢预测模型研究[J]. 工业水处理, 2025, 45(2): 132-137.
[2]	师雅琪, 孟广源, 陈鹏, 张芯婉, 付涛, 杨正武, 张连胜, 张乐华. 基于神经网络精准预测含铬废酸电氧化再生效果[J]. 工业水处理, 2025, 45(1): 131-138.
[3]	张益. 基于卷积神经网络算法的海水循环冷却污损生物分类模型[J]. 工业水处理, 2024, 44(12): 160-165.
[4]	安昱宁, 朱四富, 刘静, 杜立伟, 刘长青. 基于深度学习的污水处理厂出水总磷预测方法[J]. 工业水处理, 2024, 44(10): 143-150.
[5]	王梦妍, 王倩, 李雅婕, 刘宏波, 洪耀良. 导电材料强化厌氧处理技术研究进展[J]. 工业水处理, 2023, 43(10): 42-52.
[6]	张浩良, 刘聪, 洪乾坤, 王侃鸣, 王红宇. MBR中膜污染的人工神经网络预测研究进展[J]. 工业水处理, 2022, 42(7): 15-23.
[7]	白周央,季华,张婧伟,刘佳,杨树成. 化工厂硫酸再生冷凝液的厌氧可处理性研究[J]. 工业水处理, 2021, 41(12): 60-65.
[8]	冯胜, 刘明远, 冯旭. 基于NARX神经网络的工业水质智能处理[J]. 工业水处理, 2018, 38(3): 69-72.
[9]	张哲, 杨树成, 王慧, 冷远鹏, 李清. 铁碳微电解在造纸废水处理工艺中合理位置的探究[J]. 工业水处理, 2018, 38(2): 55-59.
[10]	张晓航, 何绪文, 王浩, 张斯宇, 孙艺欣, 李福勤, 刘涛. 神经网络预测磁絮凝处理矿井水效果的研究[J]. 工业水处理, 2018, 38(1): 57-61.
[11]	杨路, 刘惠康. 基于小波神经网络的污水处理污泥体积指数软测量[J]. 工业水处理, 2017, 37(8): 33-35.
[12]	纪振, 李东升, 王旭波, 吕文明, 李耀中. 物化预处理+厌氧+两级A/O工艺处理农药废水[J]. 工业水处理, 2017, 37(3): 104-106.
[13]	梁家豪, 刘知远, 姚显阳, 王庆宏. UASB处理含油废水的研究及微生物群落动态分析[J]. 工业水处理, 2017, 37(12): 29-33.
[14]	申明金. Kohonen-RBF网络用于废水中钴镍钒的同时测定[J]. 工业水处理, 2016, 36(8): 97-100.
[15]	张蕾, 郭宏山, 李宝忠, 陈天佐, 朱卫. 厌氧处理纤维素乙醇废水产甲烷动态研究[J]. 工业水处理, 2016, 36(10): 60-64.

选择文件类型/文献管理软件名称

选择包含的内容