杨文琚,孙晨暄,伍小龙

• 1:北京工业大学信息学部
• 2:北京工业大学计算智能与智能系统北京市重点实验室

摘要(Abstract)：

【目的】出水总氮检测在监测污水处理厂超标排放和预防水质富营养化中发挥了重要的作用，但仍存在精度低，实时性不强的问题。为解决城市污水处理过程出水总氮难以在线精准检测的问题，提出了一种基于敏感性分析的区间二型模糊神经网络(SA-IT2FNN)出水总氮软测量方法。【方法】采用敏感性分析法(Sensitivity Analysis, SA),选取与污水处理过程出水总氮关联性较强的主元变量。然后，将所选的关键主元变量作为IT2FNN的输入变量，通过训练模型参数建立出水总氮软测量模型。【结果】仿真结果显示：对于冬季数据集，SA-IT2FNN选取6个、8个、10个主元变量进行预测时，训练时间分别为6.2 s、7.8 s、9.7 s;均方根误差(Root Mean Square Error, RMSE)分别为0.44、0.35、0.33。另外，SA-IT2FNN选择8主元变量预测时，夏季和冬季对应的测试RMSE分别为0.38和0.39。【结论】结果表明：SA通过对模型的输入变量降维，有效提高了模型的预测效果；基于IT2FNN的总氮软测量模型在不同工况下都能够保证预测精度，具有较好的学习和预测能力。研究成果展示了人工智能在总氮检测中的独特作用，为污水处理出水指标高精度检测提供了有效的方法。

关键词(KeyWords)： 城市污水处理过程;出水总氮;区间二型模糊神经网络;敏感性分析法;软测量模型

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金重大项目(61890931);国家自然科学基金优秀青年项目(61622301)

作者(Author): 杨文琚,孙晨暄,伍小龙

DOI: 10.13928/j.cnki.wrahe.2023.04.011

参考文献(References)：

• [1] 姚燃，刘锋，吴露，等.三级绿狐尾藻表面流人工湿地对养殖废水处理效应研究[J].地球与环境，2018,46(5):475-481.YAO Ran,LIU Feng,WU Lu,et al.Performance of three-stage surface flow constructed wetlands planted with myriophyllum aquaticum gaudich for swine wastewater treatment[J].Earth and Environment,2018,46(5):475-481.
• [2] 张晓一，陈盛，查丽娜，等.表面流人工湿地和复合型生态浮床处理污水厂尾水的脱氮性能分析[J].环境工程，2019,37(6):46-51.ZHANG Xiaoyi,CHEN Sheng,ZHA Lina,et al.Analysis on nitrogen removal performance in treatment of wastewater treatment plant effluent with surface flow constructed wetland and integrated ecological floating-bed[J].Environmental Engineering,2019,37(6):46-51.
• [3] 薄涛.基于深度学习的污水处理厂出水总氮的预测研究[D].哈尔滨：哈尔滨工业大学，2020.BO Tao.An effluent total nitrogen prediction study of wastewater treatment plant based on deep learning[D].Harbin:Harbin Institute of Technology,2020.
• [4] 潘忠成，李敏.HJ636—2012测定总氮时影响空白值因素分析[J].环境工程，2016,34(1):126-129.PAN Zhongcheng,LI Min.Analysis of influencing factors on the blank value of total nitrogen determination by HJ 636-2012[J].Environmental Engineering,2016,34(1):126-129.
• [5] 林昆仑，秦惠，邓金花，等.一种水质总氮的检测试剂及快速检测方法：中国，CN110987918A[P],2020.LIN Kunlun,QIN Hui,DENG Jinhua,et al.A reagent and rapid detection method for total nitrogen in water:China,CN110987918A[P],2020.
• [6] 孙瑞.国内外水质总氮检测设备研究现状和发展趋势[J].山西化工，2022,42(1):42-44.SUN Rui.Research status and development trend of total nitrogen detection equipment for water quality at home and abroad[J].Shanxi Chemical Industry,2022,42(1):42-44.
• [7] LOBN A,PETER V D S,CARLOS M L V.Expanding the activated sludge model no.1 to describe filamentous bulking:The filamentous model[J].Journal of Water Process Engineering,2022(48):102896.
• [8] 韩红桂，陈治远，乔俊飞，等.基于区间二型模糊神经网络的出水氨氮软测量[J].化工学报，2017,68(3):1032-1040.HAN Honggui,CHEN Zhiyuan,QIAO Junfei,et al.Soft-sensor method of effluent ammonia nitrogen based on interval type-2 fuzzy neural networks[J].Journal of Chemical Engineering,2017,68(3):1032-1040.
• [9] ZHU S G,HAN H G,GUO M,et al.A data-derived soft-sensor method for monitoring effluent total phosphorus[J].Chinese Journal of Chemical Engineering,2017,25(12):1791-1797.
• [10] HAN H G,ZHU H G,QIAO J F,et al.Data-driven intelligent monitoring system for key variables in wastewater treatment process[J].Chinese Journal of Chemical Engineering,2018,26(10):2093-2101.
• [11] MAJID B,RECEP K D,ZHAO W Q,et al.Data to intelligence:the role of data-driven models in wastewater treatment[J].Expert Systems with Applications,2022(217):119453.
• [12] ZHAO Z H,WANG Z G,YUAN J L,et al.Development of a novel feedforward neural network model based on controllable parameters for predicting effluent Total Nitrogen[J].Engineering,2021(7):195-202.
• [13] ZIAD A G,RAMI A.Use of artificial neural network for predicting effluent quality parameters and enabling wastewater reuse for climate change resilience-a case from Jordan[J].Journal of Water Process Engineering,2021(44):102423.
• [14] 陈鹏飞，王丽学，李爱迪，等.基于灰色关联度与BP神经网络的清河水库总氮浓度预测模型[J].水电能源科学，2018,36(7):40-43.CHEN Pengfei,WANG Lixue,LI Aidi,et al.Prediction model of total nitrogen concentration in Qinghe reservoir based on grey relational grade and BP neural network[J].Hydropower Energy Science,2018,36(7):40-43.
• [15] JIANG X Q,CHEN F C,GUO L J,et al.Application of T-S Fuzzy-Neural Network Model in water quality comprehensive evaluation[J].Procedia Computer Science,2020(166):501-506.
• [16] ARMIN S B.IT2CFNN:An interval Type-2 correlation-aware fuzzy neural network to construct non-separable fuzzy rules with uncertain and adaptive shapes for nonlinear function approximation[J].Applied Soft Computing,2021(115):108258.
• [17] HAN H G,LI J M,WU X L,et al.Cooperative strategy for constructing interval type-2 fuzzy neural network[J].Neurocomputing,2019(365):249-260.
• [18] 韩红桂，鲁树武，伍小龙，等.基于改进型SVM的城市污水处理过程异常数据清洗方法[J].北京工业大学学报，2021,47(9):1011-1020.HAN Honggui,LU Shuwu,WU Xiaolong,et al.Abnormal data cleaning method for municipal wastewater treatment based on improved support vector machine[J].Journal of Beijing University of Technology,2021,47(9):1011-1020.
• [19] 何维，杨华.模型参数全局敏感性分析的EFAST方法[J].遥感技术与应用，2013,28(5):836-843.HE Wei,YANG Hua.EFAST method for global sensitivity analysis of remote sensing model′s parameters[J].Remote Sensing Technology and Application,2013,28(5):836-843.
• [20] CAO D W,LIN H C,LIU Z Y,et al.PCA-TLNN-based SERS analysis platform for label-free detection and identification of cisplatin-treated gastric cancer[J].Sensors and Actuators B:Chemical,2023(375):132903.
• [21] 陈胜勇，程徐，石凡，等.一种应用于RBF神经网络的改进Sobol敏感性分析方法：中国，CN107633294A[P],2018.CHEN Shengyong,CHENG Xu,SHI Fan,et al.An improved Sobol sensitivity analysis method applied to RBF neural network:China,CN107633294A[P],2018.
• [22] KANG W Y,SHAO B J.The impact of voice assistants′ intelligent attributes on consumer well-being:Findings from PLS-SEM and fsQCA[J].Journal of Retailing and Consumer Services,2023(70):103130.
• [23] 金珉秀.基于模糊神经网(FNN)的总氮软测量方法[D].北京：北京工业大学，2019.KIM MINSOO.A Soft-computing Method of Total Nitrogen Using Fuzzy Neural Network[D].Beijing:Beijing University of Technology,2019.