华侨大学学报(自然科学版)
華僑大學學報(自然科學版)
화교대학학보(자연과학판)
JOURNAL OF HUAQIAO UNIVERSITY(NATURAL SCIENCE)
2015年
3期
280-285
,共6页
张学阳%项雷军%林文辉%郭新华
張學暘%項雷軍%林文輝%郭新華
장학양%항뢰군%림문휘%곽신화
污水处理%神经网络%溶解氧浓度%预测控制%过程控制
汙水處理%神經網絡%溶解氧濃度%預測控製%過程控製
오수처리%신경망락%용해양농도%예측공제%과정공제
sewage treatment%neural network%dissolved oxygen concentration%predictive control%process control
针对污水处理过程溶解氧浓度时变设定值难以控制的问题,提出一种溶解氧浓度的神经网络预测控制器设计方法。首先,在活性污泥法污水处理过程通用机理模型基础上,利用系统的输入、输出数据,采用递推学习更新模式,通过三层BP神经网络训练出系统神经网络逼近模型。然后,设计满足出水水质指标的溶解氧约束预测控制器。在考虑溶解氧测量白噪音干扰和进水流量发生阶跃变化情况下,将所设计的控制器用于污水处理溶解氧浓度的时变设定值跟踪控制。仿真结果表明:与传统 PID 控制器相比,神经网络预测控制器能够显著提高溶解氧跟踪控制性能,具有更好的自适应性和抗干扰能力。
針對汙水處理過程溶解氧濃度時變設定值難以控製的問題,提齣一種溶解氧濃度的神經網絡預測控製器設計方法。首先,在活性汙泥法汙水處理過程通用機理模型基礎上,利用繫統的輸入、輸齣數據,採用遞推學習更新模式,通過三層BP神經網絡訓練齣繫統神經網絡逼近模型。然後,設計滿足齣水水質指標的溶解氧約束預測控製器。在攷慮溶解氧測量白譟音榦擾和進水流量髮生階躍變化情況下,將所設計的控製器用于汙水處理溶解氧濃度的時變設定值跟蹤控製。倣真結果錶明:與傳統 PID 控製器相比,神經網絡預測控製器能夠顯著提高溶解氧跟蹤控製性能,具有更好的自適應性和抗榦擾能力。
침대오수처리과정용해양농도시변설정치난이공제적문제,제출일충용해양농도적신경망락예측공제기설계방법。수선,재활성오니법오수처리과정통용궤리모형기출상,이용계통적수입、수출수거,채용체추학습경신모식,통과삼층BP신경망락훈련출계통신경망락핍근모형。연후,설계만족출수수질지표적용해양약속예측공제기。재고필용해양측량백조음간우화진수류량발생계약변화정황하,장소설계적공제기용우오수처리용해양농도적시변설정치근종공제。방진결과표명:여전통 PID 공제기상비,신경망락예측공제기능구현저제고용해양근종공제성능,구유경호적자괄응성화항간우능력。
In order to solve the difficult problem of controlling the dissolved oxygen (DO)concentration with time var-ying setpoint in sewage treatment process,a neural network predictive controller (NNPC)design method for the dissolved oxygen concentration is proposed in this paper.Firstly,based on the general mechanism model of the activated sludge sewage treatment process,by using the input and output data of the system and the recursive learning update mode,the neural network approximation model of the system is trained through three-layer BP neural network.Then the constrained predictive controller of dissolved oxygen is designed in the condition of satisfying effluent water quality indicators.Consid-ering the white noise interference on the dissolved oxygen measurement and the step changing influent flow,the designed controller is applied to the time varying setpoint tracking control of dissolved oxygen concentration in sewage treatment process.Simulation results show that compared to the conventional PID controller,the neural network predictive control-ler can significantly improve the tracking control performance of dissolved oxygen concentration and has better adaptability and stronger disturbance rej ection ability.