﻿ 基于ESN的污水处理过程优化控制
 文章快速检索 高级检索

Optimal control for wastewater treatment process based on ESN neural network
QIAO Junfei, WANG Lili, HAN Honggui
College of Electronic and Control Engineering, Beijing University of Technology, Beijing 100124, China
Abstract: To address the problem of the high-energy consumption in wastewater treatment processes, we propose an online optimal control method based on an echo state network(ESN). This method has three main steps. First, we develop a performance prediction model of the wastewater treatment process. Second, based on the system state and the predicted performance index, we optimize the set point of the control variable using the ESN in real time. Then we transfer the optimized set point to the underlying controller for tracking control. This ESN-based online optimal control method is carried out using the benchmark simulation model 1(BSM1). The simulation results show that the proposed method can not only meet the effluent quality requirements, but also efficiently reduce the operation costs of the wastewater treatment process.
Key words: wastewater treatment process     optimal control     echo state network     performance prediction model     benchmark simulation model

Vrbie等[12]采用策略迭代的方法，通过对当前控制策略的性能评估来对控制策略进行改善，并指出一个智能控制器应该是可以根据外界环境的变化而动态改变控制策略的。文中设计了一种自适应在线优化控制器，通过神经网络预测未来的性能指标函数值，然后将其传递给神经网络优化层，根据系统的环境状态以及上一时刻的控制变量设定值来优化预测的指标函数值，使其达到最小，从而产生新的设定值送到底层控制器进行跟踪，实现污水处理过程的优化控制。

1 污水处理过程模型

 图 1 BSM1的设备布局图 Fig. 1 The equipment layout of BSM1

2 基于ESN的污水处理优化控制 2.1 优化问题描述

Brdys等[5]指出污水处理厂的控制意义在于保证其稳定运行，在使出水水质达标的同时尽量减少能耗。因此，污水处理过程优化是一个多目标优化的过程，不仅要考虑出水水质是否达标，还要考虑是否能节省运行费用。基于以上，文中定义优化的性能指标函数如式(6)所示：

 系数 B1 B2 B3 B4 B5 值 2 1 10 30 2

1)物料平衡的约束：

2)出水水质约束：

3)执行器约束：

2.2 优化控制系统

 图 2 优化控制系统结构图 Fig. 2 The structure of optimal control system
2.2.1 性能指标预测模型

2.2.2 神经网络优化模型

ESN的动态储备池输出为

3 实验研究

 图 3 溶解氧优化效果 Fig. 3 Optimization results of the dissolved oxygen concentration
 图 4 硝态氮优化效果 Fig. 4 Optimization results of the nitrate concentration
 图 5 出水BOD的效果比较 Fig. 5 Comparison results of effluent BOD
 图 6 出水COD的效果比较 Fig. 6 Comparison results of effluent COD
 图 7 出水TSS的效果比较 Fig. 7 Comparison results of effluent TSS
 图 8 出水氨氮的效果比较 Fig. 8 Comparison results of effluent SNH
 图 9 出水总氮的效果比较 Fig. 9 Comparison results of effluent Ntot

 控制策略 BOD5/gCOD·m-3 COD/gCOD·m-3 SNH/gN·m-3 Ntot/gN·m-3 TSS/gSS·m-3 闭环控制 2.676 3 47.511 4 2.303 7 16.800 6 12.621 9 优化控制 2.678 4 47.523 5 2.872 3 15.601 4 12.591 7

 控制策略 EQ/kgpoll.units·d-1 AE/kWh·d-1 PE/kWh·d-1 Energy/kWh·d-1 闭环控制 6 080.9 3 677.016 232.455 3 909.5 优化控制 6 134.5 3 494.7 262.130 8 3 756.8
5 结束语

 [1] HAIMIA H, MULAS M, CORONA F, et al. Data-derived soft-sensors for biological wastewater treatment plants:an overview[J]. Environmental Modelling and Software, 2013, 47:88-107. [2] ZHANG Ping, YUAN Mingzhe, WANG Hong. Improvement of nitrogen removal and reduction of operating costs in an activated sludge process with feedforward-cascade control strategy[J]. Biochemical Engineering Journal, 2008, 41(1):53-58. [3] OSTACE G S, CRISTEA V M, AGACHI P S. Cost reduction of the wastewater treatment plant operation by MPC based on modified ASM1 with two-step nitrification/denitrification model[J]. Computers and Chemical Engineering, 2011, 35(11):2469-2479. [4] PIOTROWSKI R, BRDYS M A, KONARCZAK K, et al. Hierarchical dissolved oxygen control for activated sludge processes[J]. Control Engineering Practice, 2008, 16(1):114-131. [5] BRDYS M A, GROCHOWSKI M, GMINSKI T, et al. Hierarchical predictive control of integrated wastewater treatment systems[J]. Control Engineering Practice, 2008, 16(6):751-767. [6] HOLENDA B, DOMOKOS E, RÉDEY Á, et al. Dissolved oxygen control of the activated sludge wastewater treatment process using model predictive control[J]. Computers and Chemical Engineering, 2008, 32(6):1270-1278. [7] 薄迎春, 乔俊飞. 启发式动态规划在污水处理过程控制中的应用[J]. 控制理论与应用, 2013, 30(7):828-833. BO Yingchun, QIAO Junfei. Application of heuristic dynamic programming to wastewater treatment process control[J]. Control Theory and Applications, 2013, 30(7):828-833. [8] 丛秋梅, 柴天佑, 余文, 等. 污水处理过程的递阶神经网络建模[J]. 控制理论与应用, 2009, 26(1):8-14. CONG Qiumei, CHAI Tianyou, YU Wen, et al. Modeling wastewater treatment plant via hierarchical neural networks[J]. Control Theory and Applications, 2009, 26(1):8-14. [9] JOUANNEAU S, RECOULES L, DURAND M J, et al. Methods for assessing biochemical oxygen demand(BOD):a review[J]. Water Research, 2014, 49:62-82. [10] IQBAL J, GURIA C. Optimization of an operating domestic wastewater treatment plant using elitist non-dominated sorting genetic algorithm[J]. Chemical Engineering Research and Design, 2009, 87(11):1481-1496. [11] 史雄伟, 乔俊飞, 苑明哲. 基于改进粒子群优化算法的污水处理过程优化控制[J]. 信息与控制, 2011, 40(5):698-703. SHI Xiongwei, QIAO Junfei, YUAN Mingzhe. Optimal control for wastewater treatment process based on improved particle swarm optimization algorithm[J]. Information and Control, 2011, 40(5):698-703. [12] VRABIE D, LEWIS F. Neural network approach to continuous-time direct adaptive optimal control for partially unknown nonlinear systems[J]. Neural Networks, 2009, 22(3):237-246. [13] COPP J B. The COST simulation benchmark:description and simulator manual[R]. Luxembourg:Office for Publications of the European Community, 2001. [14] QIAO Junfei, BO Yingchun, CHAI Wei, et al. Adaptive optimal control for a wastewater treatment plant based on a data-driven method[J]. Water Science and Technology, 2013, 67(10):2314-2320. [15] CHACHUAT B, ROCHE N, LATIFI M. Long-term optimal aeration strategies for small-size alternating activated sludge treatment plants[J]. Chemical Engineering and Processing:Process Intensification, 2005, 44(5):591-604.
DOI: 10.11992/tis.201401009

0

#### 文章信息

QIAO Junfei, WANG Lili, HAN Honggui

Optimal control for wastewater treatment process based on ESN neural network

CAAI Transactions on Intelligent Systems, 2015, 10(6): 831-837.
DOI: 10.11992/tis.201401009