改进K-means算法优化RBF神经网络的出水氨氮预测  

Prediction of Effluent Ammonia Nitrogen Based on Improved K-means Algorithm Optimizing RBF Neural Network

作　　者：乔俊飞 孙玉庆 韩红桂 

机构地区：[1]北京工业大学信息学部

出　　处：《控制工程》2018年第3期375-379,共5页Control Engineering of China

基　　金：国家自然科学基金(61203099,61225016);北京市科技计划课题(Z141100001414005,Z141101004414058);中国博士后科学基金资助项目(2014M550017,XJ2013018);北京市科技新星计划(Z131104000413007);教育部博士点基金项目(20121103120020,20131103110016);北京市教委项目(km201410005001,KZ201410005002);北京市朝阳区博士后资助项目(2014ZZ-05);北京市朝阳区协同创新项目(ZH14000177)

摘　　要：为提高污水处理过程中出水氨氮的预测精度，并针对RBF神经网络参数难以确定的问题，提出一种改进K-means算法优化RBF神经网络的氨氮预测算法。首先，计算每个样本点的密度值，以其大小是否满足一个阈值为条件，判定该点是否为孤立点或噪声点，来消除孤立点和噪声点对K—means算法的影响;然后利用减法聚类算法初始化K—means算法的聚类中心，并得到聚类中心的个数，将改进后的K-means算法优化RBF神经网络结构；最后，通过对污水处理过程中出水氨氮的实际预测实验，表明所提出的算法具有较强的逼近能力。In order to improve the prediction accuracy of ammonia nitrogen in wastewater, and solve the problems that the parameters of RBF neural network are difficult to determine, this paper puts forward a prediction algorithm of ammonia nitrogen based on improved K-means algorithm optimizing RBF neural network. Firstly, calculate the density of each sample point, and determine whether the point is an isolated point or a noise point based on whether the value is satisfied by a threshold value or not, so as to eliminate the effect of outliers and noise points on the K-means algorithm; Then the subtractive clustering algorithm is used to initialize the clustering centers of the K-means algorithm, meanwhile the number of cluster centers is gotten, and the improved K-means algorithm is taken to optimize the structure of RBF neural network; Finally, the experiment of prediction of ammonia nitrogen in wastewater is done to indicate that the proposed algorithm has strong approximation ability.

关 键 词：氨氮预测 RBF神经网络 K-MEANS算法 密度指标 

分 类 号：TP173[自动化与计算机技术—控制理论与控制工程]