改进型T-S模糊神经网络的出水氨氮预测  

Prediction of Ammonia Nitrogen in Effluent By iImproved T-S fuzzy neural network

作　　者：崔心惠 李文萱[1] 詹玉新 CuiXinhui;Li Wenxuan;Zhan Yuxin(College of Electrical Engineering,Chuzhou Polytechnic,Chuzhou 239000,China)

机构地区：[1]滁州职业技术学院电气工程学院,安徽滁州239000

出　　处：《云南化工》2021年第8期104-107,共4页Yunnan Chemical Technology

基　　金：2020年度安徽省高校优秀青年人才支持计划重点项目(gxyqZD2020068);2020年校级科研一般项目(YJY-2020-25)。

摘　　要：由于污水在复杂的处理过程中无法进行出水氨氮(NH_(4)-N)值的实时检测,提出一种改进型T-S模糊神经网络(TSFNN)的算法来对出水NH4-N进行有效预测。利用主元分析法和"统计产品与服务解决方案"(SPSS)软件对出水NH4-N浓度的关键影响因子进行筛选,选择主要特征变量作为输入。针对TSFNN结构,采用模糊C-均值聚类算法(FCM)对其进行初始化。结果表明,该算法对出水氨氮值的预测具有良好的预测精度且保证了处理过程的时效性。Because the effluent ammonia nitrogen( NH_(4)-N) can not be detected in real time in the complex wastewater treatment process,an improved T-S fuzzy neural network( TSFNN) algorithm is proposed to effectively predict the effluent NH_(4)-N. Firstly,principal component analysis( PCA) and statistical products and services solutions( SPSS) software were used to screen the key influencing factors of effluent NH_(4)-N concentration,and the main characteristic variables were selected as input. For TSFNN structure,fuzzy c-means clustering algorithm( FCM) is used to initialize it. The simulation results show that the algorithm has good prediction accuracy for effluent ammonia nitrogen value and ensures the timeliness of the treatment process.

关 键 词：出水氨氮 模糊C-均值聚类 T-S模糊神经网络 预测 

分 类 号：TP15[自动化与计算机技术—控制理论与控制工程]