基于VW-IGRBF神经网络的出水BOD软测量

doi:10.16180/j.cnki.issn1007-7820.2022.05.005

摘要/Abstract

摘要：

针对污水处理过程具有复杂非线性特性以及出水BOD难以精确测量的问题,文中提出一种基于变宽度的逆平方根和高斯函数线性组合的RBF神经网络软测量方法。神经网络的激活函数由逆平方根函数和高斯函数线性组合,弥补了单一激活函数在某些区间饱和的问题,提高了隐层激活函数的表达能力和自适应能力。由于激活函数的宽度对模型的泛化性能有较大的影响,因此引入基于核密度的变宽度策略可以有效提高网络泛化能力。文中采用改进LM算法实现了神经网络参数的在线学习。基于污水处理过程实际运行数据的仿真实验表明,所提方法对于出水BOD具有较高的预测精度和良好的自适应能力。

关键词: 变宽度, 组合函数, RBF, 软测量, 出水BOD, 激活函数, 核密度, LM算法

Abstract:

In view of the problem that the wastewater treatment process has complex nonlinear characteristics and the effluent BOD is difficult to accurately measure, a soft measurement method based on VW-IGRBF neural network is proposed in this study. The activation function of the neural network is a linear combination of the inverse square root function and the Gaussian function, which makes up for the saturation of a single activation function in certain intervals, and improves the expression and self-adaptability of the hidden activation function. Since the width of the activation function has a greater impact on the generalization performance of the model, a variable width strategy based on kernel density is introduced to improve the generalization ability of the network. In this study, the improved LM algorithm is used to realize the online learning of neural network parameters. Simulation experiments based on actual operating data of the wastewater treatment process show that the proposed VW-IGRBF method has higher prediction accuracy and better adaptive ability for effluent BOD.

Key words: variable width, combinational functionn, RBF, soft-sensing, effluent BOD, activation function, kernel density, LM algorithm

中图分类号:

TP183

赵豆豆,张伟. 基于VW-IGRBF神经网络的出水BOD软测量[J]. 电子科技, 2022, 35(5): 26-32.

ZHAO Doudou,ZHANG Wei. Soft-Sensing of Effluent BOD Based on VW-IGRBF Neural Network[J]. Electronic Science and Technology, 2022, 35(5): 26-32.

图/表 10

表1

图1

图2

图3

图4

图5

图6

图7

表2

表3

参考文献 21

[1]	Heddam S, Lamda H, Filali S. Predicting effluent biochemical oxygen demand in a wastewater treatment plant using generalized regression neural network based approach: a comparative study[J]. Environmental Processes, 2016, 3(1):153-165. doi: 10.1007/s40710-016-0129-3
[2]	丁海旭, 李文静, 叶旭东, 等. 基于自组织递归模糊神经网络的BOD软测量[J]. 计算机与应用化学, 2019, 36(4):331-336.
	Ding Haixu, Li Wenjing, Ye Xudong, et al. BOD soft-sensing based on self-organizing recursive fuzzy neural network[J]. Computer and Applied Chemistry, 2019, 36(4):331-336.
[3]	柴伟, 纪镐南. 污水处理出水BOD区间预测建模[J]. 哈尔滨工业大学学报, 2018, 50(2):71-76.
	Chai Wei, Ji Haonan. Interval predictor models for effluent BOD of wastewater treatment[J]. Journal of Harbin Institute of Technology, 2018, 50(2):71-76.
[4]	Yan A, Shao H, Wang P. A soft-sensing method of dissolved oxygen concentration by group genetic case-based reasoning withintegrating group decision making[J]. Neurocomputing, 2015, 16(9):422-429.
[5]	乔俊飞, 李瑞祥, 柴伟, 等. 基于PSO-ESN神经网络的污水BOD预测[J]. 控制工程, 2016, 23(4):463-467.
	Qiao Junfei, Li Ruixiang, Chai Wei, et al. Prediction of sewage BOD based on PSO-ESN neural network[J]. Control Engineering of China, 2016, 23(4):463-467.
[6]	Song C M, Kim J S. Applicability evaluation of the hydrological image and convolution neural network for prediction of the biochemical oxygen demand and total phosphorus loads in agricultural areas[J]. Agriculture, 2020, 10(11),529-536. doi: 10.3390/agriculture10110529
[7]	鲁明, 刘东儒, 章毅, 等. 基于灰狼算法与RBF神经网络对污水出水BOD的预测[J]. 兰州工业学院学报, 2019, 26(2):83-87.
	Lu Ming, Liu Dongru, Zhang Yi, et al. Prediction of sewage water BOD based on improved grey wolf algorithm and RBF neural network[J]. Journal of Lanzhou Institute of Technology, 2019, 26(2):83-87.
[8]	Meng X, Rozycki P, Qiao J F. Nonlinear system modeling using RBF networks for industrial application[J]. IEEE Transactions on Industrial Informatics, 2018, 14(3):931-940. doi: 10.1109/TII.2017.2734686
[9]	乔俊飞, 马士杰, 杨翠丽. 基于ROLS算法的递归RBF神经网络结构设计[J]. 化工学报, 2018, 69(3):1191-1199.
	Qiao Junfei, Ma Shijie, Yang Cuili. Structure design for recurrent RBF neural network based on recursive orthogonal least squares[J]. CIESC Journal, 2018, 69(3):1191-1199.
[10]	苑仁令, 向凤红, 毛剑琳, 等. 基于视觉检测板球系统的RBF-PID控制研究[J]. 电子科技, 2018, 31(5):23-27.
	Yuan Renling, Xiang Fenghong, Mao Jianlin, et al. Research on RBF-PID Control based on visual Inspection Cricket system[J]. Electronic Science and Technology, 2018, 31(5):23-27.
[11]	Gholampour F, Hesameddini E, Taleei A. A stable RBF partition of unity local method for elliptic interface problems in two dimensions[J]. Engineering Analysis with Boundary Elements, 2021, 12(3):220-232. doi: 10.1016/0955-7997(93)90019-H
[12]	Kacha A, Grenez F, Orozco-Arroyave J R. Principal component analysis of the spectrogram of the speech signal: Interpretation and application to dysarthric speech[J]. Computer Speech and Language, 2020, 59(2):114-122. doi: 10.1016/j.csl.2019.07.001
[13]	翟超, 熊伟丽. 一种分层信息提取的多块主元分析故障监测方法[J]. 南京理工大学学报, 2020, 44(4):471-480.
	Zhai Chao, Xiong Weili. Multi-block fault monitoring based on PCA method with hierarchical information extraction[J]. Journal of Nanjing University of Science and Technology, 2020, 44(4):471-480.
[14]	Khan S, Naseem I, Togneri R, et al. A novel adaptive kernel for the RBF neural networks[J]. Circuits,Systems,and Signal Processing, 2017, 36(4):1639-1653. doi: 10.1007/s00034-016-0375-7
[15]	黄昊, 胡永峰. 基于采样点局部密度的径向基函数核宽度确定方法及其在装备设计中的应用范例[J]. 军事运筹与系统工程, 2016, 30(4):58-64.
	Huang Hao, Hu Yongfeng.Radial basis function kernel width determination method based on sampling point local density and its application in equipment design[J]. Military Operations Research and SystemsEngineering, 2016, 30(4):58-64.
[16]	Erelik E, Nadar M. Nonparametric density estimation based on beta prime kernel[J]. Communications in Statistics-Theory and Methods, 2020, 49(2):325-342. doi: 10.1080/03610926.2018.1538458
[17]	Cui H, Xia G, Jin S Q, et al. Levenberg-Marquardt algorithm with adaptive Tikhonov regularization for bandwidth correction of spectra[J]. Journal of Modern Optics, 2020, 67(7):661-670. doi: 10.1080/09500340.2020.1766590
[18]	王仁超, 吴松. 基于LM算法的神经网络模型预测爆破块度[J]. 水力发电学报, 2019, 38(7):100-109.
	Wang Renchao, Wu Song. Neural network model based prediction of fragmentation of blasting using theLevenberg-Marquardt Algorithm[J]. Journal of Hydroelectric Engineering, 2019, 38(7):100-109.
[19]	张昭昭, 乔俊飞, 余文. 基于LM算法的在线自适应RBF网结构优化算法[J]. 控制与决策, 2017, 32(7):1247-1252.
	Zhang Zhaozhao, Qiao Junfei, Yu Wen. Online self-adaptive optimal algorithm for RBF network based on Levenberg-Marquardt algorithm[J]. Control and Decision, 2017, 32(7):1247-1252.
[20]	乔俊飞, 李微, 韩红桂. 改进的T-S模糊神经网络用于生化需氧量的软计算[J]. Chinese Journal of Chemical Engineering, 2014, 22(Z1):1254-1259. doi: 10.1016/j.cjche.2014.09.023
	Qiao Junfei, Li Wei, Han Honggui. Soft computing of biochemical oxygen demand usingan improved T-S fuzzy neural network[J]. Chinese Journal of Chemical Engineering, 2014, 22(Z1):1254-1259. doi: 10.1016/j.cjche.2014.09.023
[21]	李文静, 李萌, 乔俊飞. 基于互信息和自组织RBF神经网络的出水BOD软测量方法[J]. 化工学报, 2019, 70(2):687-695.
	Li Wenjing, Li Meng, Qiao Junfei. Effluent BOD soft measurement based on mutual information and self-organizing RBF neural network[J]. CIESC Journal, 2019, 70(2):687-695.

序号	辅助变量	简称
1	出水总氮	TN₁
2	出水氨氮	NH₃-N₁
3	进水总氮	TN₂
4	进水生物需氧量	BOD
5	进水氨氮	NH₃-N₂
6	出水磷酸盐	PHOS₁
7	生化混合液悬浮固体	MLSS
8	生化池溶解氧	DO
9	进水磷酸盐	PHOS₂
10	进水化学需氧量	COD

算法	方差	标准差	MAE
RBFNN	0.387 0	0.622 1	0.476 7
IGRBFNN	0.360 8	0.604 5	0.491 4
VW-IGRBFNN	0.313 0	0.559 3	0.445 2

算法	测试RMSE	测试MSE
FS-RBFNN^[9]	0.549 1	0.301 5
TS-FNN^[20]	0.420 7	0.177 0
ESRBFNN-MI^[21]	0.388 3	0.150 8
VW-IGRBFNN	0.233 7	0.054 6