节点文献

城市供水水处理系统的建模、控制与运行优化研究

Research on Modeling and Control and Operation Optimization of Water Treatment System for Urban Water Supply

【作者】 唐德翠

【导师】 朱学峰;

【作者基本信息】 华南理工大学 , 模式识别与智能系统, 2013, 博士

【摘要】 城市供水水处理系统作为一个城市的生活基础设施,其供水质量关系到居民的身体健康,混凝沉淀和过滤作为水处理系统的关键工序,其中水的混凝效果及滤池工况既决定出水质量还影响制水成本。因此,对水处理系统建模,控制与运行优化进行深入研究,在保证供水质量安全的同时尽可能降低能耗是十分有意义的研究课题。在相关科研项目的支持和广东某水厂的协助配合下,对水处理系统的建模、控制与运行优化进行了四方面的研究,主要研究内容和取得成果如下:1.对混凝投药进行了建模研究,建立了三种不同结构的数学模型。针对目前国内绝大部分水厂投药量由人工经验确定的不足,建立了一种简化适用的投药指数模型,提出了一种改进的差分进化算法,并用该算法来辨识指数模型中的参数;结合现有水厂的实际投药方法,建立了一种投药分段线性模型,该模型以源水浊度变化分区间,以源水浊度、源水流量为前馈量实现投药粗调,以待滤水浊度为反馈量实现投药细调;最后建立了一种性能较优的神经网络模型。各自的模型检验结果表明得到的三种模型较为准确,可利用其计算结果指导现场操作人员投药,减少生产过程投药的盲目性,确保了待滤水浊度的稳定。2.对混凝沉淀过程的控制算法进行了分析,研究了几种控制算法并对其做了仿真分析。针对对象模型不易获取,提出了两种改进的PID算法:基于改进差分辨识的PID(MDE-PID)和基于迭代反馈的二自由度PID(IFT-PID)。MDE-PID算法中采用了一种根据个体适应值优劣来变异的改进策略,仿真结果表明该算法可以提高收敛速度和收敛精度。IFT-PID算法对迭代步长更新作了适当改进以确保获得解的准确性,仿真结果表明该算法辨识得到解比其它传统方法更优。针对对象受扰因素多和大滞后特性,提出了两种新的控制算法:基于Smith模型预估的双控制器控制和基于迭代反馈的内模控制(IFT-IMC)。Smith模型预估的双控制器方案将Smith的模型补偿优点和双回路的独立性控制优点结合起来,并对跟踪PI控制器进行非线性补偿。仿真结果表明此算法具有较好的自适应能力和稳定性,可实现对大滞后、干扰多的复杂系统的良好控制。IFT-IMC算法集合了内模控制鲁棒性强和IFT自适应强的优点,具有较好的自适应性和鲁棒性,控制性能优于IMC-PID和Smith-PID,可用于实现对此类难控系统的有效控制。3.对滤池水头损失和比沉积变化进行了实验研究,建立了滤池水头损失和比沉积两种数学模型。针对过滤过程中水头损失不易求取和比沉积无法获得的不足,建立了水头损失数学模型和比沉积数学模型。首先根据实验数据求取了不同滤层深度水头损失值,给出了水头损失随滤层深度变化的指数关系式以及一种新的水头损失与比沉积的定量关系式,求取了不同滤层深度比沉积。然后借助获得的关系和现有文献结果,给出了水头损失和比沉积随各种影响因素之间的关系式。最后借助获取的水头损失数据和比沉积数据,采用改进差分方法辨识各自模型参数,建立了水头损失模型和比沉积模型,各自的模型检验结果表明,求得的两种模型是准确的。借助两种模型可以从水头损失宏观角度和比沉积微观角度了解滤层堵塞情况,从而更好的设置过滤周期。4.建立了最优待滤水浊度数学模型,并对滤池过滤和反冲洗进行了运行优化实验研究。从总能耗角度出发,给出了能耗最低时的最优待滤水浊度具体表达式,采用回归辨识法获得了表达式中的未知参数,从而建立了最优待滤水浊度数学模型,证明了现有研究成果中定性提出的最优经济浊度概念的存在性。分析了课题水厂滤池的目前运行工况,提出了两种优化实验方案,并对滤池优化前后的水头损失做了对比分析,实验结果表明,本文提出的优化结论是合理的,实施该优化方案,可以比较明显地降低能耗,具有较好的经济效益和社会效益。最后在总结本文研究的基础上,对水处理系统中建模与控制技术的今后发展及联合优化运行方面进行了展望。

【Abstract】 City water supply as a city infrastructure,the water quality of water treatment system isclosely related to the people’s health.As the primary part of the water treatment process, thecoagulation result of water will influence the subsequent processing units,filtration as afollow-up treatment process,it determines the quality and the cost of water.Therefore, toensure water quality and safety while minimizing energy consumption,modeling, control andoperation optimization of the water treatment system in-depth study is very meaningfulresearch.with the support of the relevant research projects and the cooperation of the relevantwaterworks,the study on water treatment system includes4aspects, and the main contentsand research achievements are as follows:1. Study the modeling of the dosage in the coagulation process, and establish three differentstructural mathematical models.Aiming at the arbitrariness of dosage usage in current water treatment process, thereduced exponential model of coagulation dosage has been established based on existingresearch achievement and the realtime data, the parameters of the exponential model havebeen identified by using an improved differential evolution algorithm. Depending on thescope of raw water turbidity, the piecewise linear model has been established, which includesthree input parameters:the raw water turbidity, raw water flow and the pending filter turbidity,the former two as feed-forward variable adjustment and the latter one as feedback variableadjustment.the parameters of the piecewise linear model have been identified by using linearregression analysis. Finally using neural network technology to model, a neural networkblack model with the best performance has been given by comparing and analysingperformance of model,which is composed of different intermediate nodes and mappingfunctions. The model test results show that the models are accurate and instructive, accordingto the model to achieve accurate control of dosage, reduce blindness in the production processand ensure the stability of water quality.2. Analysis the control algorithms of the coagulation process in-depth and study severalcontrol algorithms and analysis their simulation results. Aiming at the difficulty to obtain the model of coagulation process, two improved PIDalgorithm have been introuuced:the PID controller based on improved differentialevolution(MDE-PID) and the two freedom degrees PID based on iterative feedback(IFT-PID).MDE-PID algorithm uses an improved mutation strategy that the target vector depends on thebest vector according to their fitness value, the simulation results show that the algorithm canimprove the convergence speed and accuracy. for the IFT-PID algorithm, the effects of theiterative step and the parameters’ initial value on the iterative process have been discussedand the iterative step has been improved,the simulation results show that the method is betterthan other traditional methods.Taking into consideration more disturbance and large delay,two new algorithms have been proposed: dual controllers scheme based on Smith modelcontrol and internal model control based on iterative feedback tuning(IFT-IMC). The dualcontrollers scheme combines the advantage of the traditional Smith predictor and theindependence of dual controllers,and compensate the tracking PI controller. the simulationresults show that the adaptation to object parameters change and the stability are better.IFT-IMC algorithm combines the strong robustness of internal model control and the strongadaptivity of IFT, it has good adaptability and robustness and the control performance is bestamong IMC-PID and Smith-PID, so it can be used to control this system to achieve effectivecontrol.3. Study on the change of the headloss and sludge content per unit volume during filtrationprocess from the experimental point of view and establish two kind of mathematical modelsfor filter.Aiming at the difficulty to obtain the headloss and sludge content per unit volume duringfiltration process, the mathematical models of head loss and sludge content per unit volumehave been established. Firstly, the headloss values of different filter depth have been obtainedby means of experiment data, the exponential relationship of headloss with the filterlayer depth change and the quantitative relation between the headloss and sludge content aregiven, so the sludge content of different layer depth has been calculated. Then with the helpof the relations and the existing research results,the relationship of the headloss with thevarious influencing factors and the relationship of the sludge content with thevarious influencing factors have been obtained. Finally, based on the data of headloss and sludge content, adopted modified differential method to identify the parameters,the model ofheadloss and sludge content are established and the model test results shows that the twomodels are accurate respectively. With the help of two kind of models, from themacro perspective of headloss and the micro perspective of sludge content, the cloggedsituation of the filter layer is clear and the filter cycle has been set more reasonabley.4. Establish the turbidity mathematical model of water pending treatment and study onoptimization experiment of filter filtration and backwash.From the perspective of the total energy consumption, the energy consumption equationof coagulation and filtration has been made,so the optimal turbidity expression has beenobtained when the total energy consumption is minimal. Based on the filter parameters andreal-time data, the expression parameters has been identified by using recursive identificationmethod.this research result proves the existence of optimal economic turbidity concept thathas been proposed qualitatively.Then the actual work situation of water plant has beenanalyzed and the optimization experimental study on the filtration cycle and backwashingtime has been implemented,lastly comparison of energy consumption between before andafter optimization has been discussed, The experimental results show that optimizing theexisting filter conditions appropriatly can reduce energy consumption significantly andsave the cost of water, which has good economic benefit and social benefit.Finally, the research achievements of this paper have been concluded and the future ofdevelopment of modeling,control method and joint optimal operation has been introduced.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络