【作者】 郑爽；

【导师】 李鱼；

【作者基本信息】 吉林大学 ， 环境科学， 2010， 博士

【摘要】 催化湿式氧化法是一种处理高浓度有机废水的有效方法,其处理效果主要受原液浓度、催化剂特性、反应温度、反应时间、氧分压等因素的影响。Mn-Ce催化剂在很多催化湿式氧化反应中显现出很高的催化活性,催化剂制备过程中影响催化剂活性的主要因素包括焙烧温度、焙烧时间及金属配比等。本文利用多元非线性回归和人工神经网络方法对Mn-Ce催化剂的制备及催化湿式氧化降解正丁酸的升温、恒温过程分别进行模拟,并综合考虑实验成本及降解效果约束,结合相应的优化算法对反应条件进行优化。对比发现,在Mn-Ce催化剂制备及催化湿式氧化的恒温反应过程中,人工神经网络相比非线性回归方法优化效果更理想,并将其应用到间硝基苯磺酸钠催化湿式氧化降解条件优化的研究中:带成本约束的降解条件优化无论是从降解效果还是成本上都起到很好的优化作用。人工神经网络及非线性回归方法均可以帮助实践较快寻求到经济性的最优条件,起到减少浪费、节省时间的作用,对指导实践有着重要的意义。更多还原

【Abstract】 With the development of economy and the improvement of the urbanization level and people’s living standards, municipal waste production was growing, and their physical and chemical composition as well as the nutritional characteristics was taking place with the significant changes of economic development and people’s living standards. At present, the methods of handling municipal solid waste including sanitary landfill disposal, high-temperature composting, incineration and the integrated treatment of refuse resources. More than 70% China’s urban waste was disposed by sanitary landfill. Leachate was poisonous waste water containing high amounts of organics, and even leachate was also regarded as non-degradation organic compound and hard to be treated by most of the common disposal methods. Landfill leachates may be possible to pollute the surface water, ground water and the substrate of the landfill site. The landfill leachates were generated from precipitation, rainfall and the water in wastes. The organic compounds contained in landfill leachates were much more complicated than other aqueous. Besides that the level of COD could reach 7.0×10⁴ mg/L, the value of BOD₅ was 3.8×10⁴ mg/L, and the concentration of the ammonia nitrogen （NH₄⁺-N） reached 1.7×10³ mg/L, the landfill leachates contained high concentrations of many heavy metals such as Fe, Pb and so on. The main organic chemicals such as alkanes, alkenes, aromatic hydrocarbons, organic acids, lipids, carbonyl compounds and alcohols could be detected in landfill leachates as well. Due to its serious pollution to the surface water, ground water and the soil of the landfill sites, the long-term collection of the landfill leachates and the secondary pollution were easy to be occurred. Therefore, it is an duty to collect the landfill leachates properly and effectively treat the wastewater - leachates, especially around cities no matter the home and abroad. The treatment technologies used to dispose leachates were hot topics on earth. The main methods, for example physical-chemical methods, biological methods,and the combination of several different methods were usually employed to process landfill leachates. Biological methods had many advantages such as easy to be operated, lower the run cost, and the well developed operation technique. Hence, biological methods were popular and applied extensively. Some waste water was known to be of low-biodegradability, resistace to degradation and poisonous to microbial, which was not suitable to be disposed using biological methods. Physical-chemical methods had some unique merits, for example, the commonly employed technics, including photocatalysis oxidation, fenton oxidation, and physical adsorption, membrane separation, chemical precipitation, electrolytic oxidation method and other advanced oxidation technologies such as catalytic wet air oxidation （CWAO） and so on.CWAO was also to be considered as a kind of physical-chemical method, which was properly to be used to treat the high concentrated organic wastewater, and not properly to process those wastewater which contained too small amount of organic compounds to incinerate and too high levels of organic compounds to biological treatment. CWAO could be applied broadly to be in pratice because the treatment process was time-saving, the removal efficiency was higher, the treatment cost was lower and the chance of second pollution risk was reduce significantly. Since CWAO had been developed, the method was concerned by researchers. Under the conditions of high pressure of oxygen and temperatures, CWAO was conducted to dispose organics in wastewater with the help of some catalysts. Recently, many satisfying experimental results which employed the method of CWAO to dispose many styles of wastewater were obtained by many researchers in home and abroad. During the experiment, the degradation effecicy were affected by catalyst used, time, temperature, the levels of organic compounds and oxygen pressure. Results of degradation varied with the different degradation conditions. In order to obtain high efficiency and more economical to conduct the experiment, therefore, it was important to select the experimental conditions.Meanwhile, Mn-Ce catalyst shows a high catalytic activity in lots of CWAO reactions, and it has a great influence of catalytic activity and butyric acid degradation results with different catalyst preparation conditions. Therefore, it is significance for selecting a catalyst with higher catalytic activity and improving the CWAO degradation efficiency by researching the degradation effect with different catalyst preparation conditions.With the development of science and technology, a number of cross-science was gradually formed with the trends of "mathematical" and "intelligent". Mn-Ce catalyst preparation and CWAO degradation reaction are nonlinear dynamic processes; there are two main methods for establishing a non-linear mapping relation: non-linear regression method and artificial neural network model. The experimental data was organized and analysised by the mathematics basic theory and computational methods in the study of Mn-Ce catalyst preparation conditions and butyric acid CWAO degradation process through non-linear regression method, it is great helpful for Mn-Ce catalyst preparation conditions and butyric acid CWAO degradation mathematical simulation process by the investigation of mutual relationships between the various factors and the prediction of trends. Artificial neural network is an intelligent simulation model based on physiology; it is connected by a large number of processing units which composed of a large-scale adaptive non-linear dynamic system; it is often used to fitting and predicting complex nonlinear systems with its characteristics of strong self-learning, self-adaptive, anti-jamming, et al. In the simulation process of Mn-Ce catalyst preparation conditions and butyric acid CWAO degradation, the method of predicting the unknown results by finding out the mapping relations through the study of known experimental data has a quite broadly application prospects.In this paper, the influences of different Mn-Ce catalyst preparation conditions and different degradation reaction conditions during the heating and constant temperature processes on the catalytic activity prepared （expressed by TOC removal rate） and the butyric acid degradation effect （expressed by TOC removal rate） was studied. Non-linear regression method and artificial neural network model were established for different processes based on experimental data, and optimized the preparation and degradation conditions considered the experimental cost and degradation effect. In addition, it has proved that the artificial neural network method was well applied in the sodium metanitrobenzene sulfonate degradated by CWAO.Firstly, multiple non-linear regression mathematical model and BP neural network method were established to investigate the relationships between three main factors （baking temperature, baking time, metal matching） and butyric acid degradation result respectively, according to the experimental data of Mn-Ce catalyst preparation. Validated the models reliability by error analysis, and Matlab multi-variable optimization function--Fmincon was employed to optimize the experimental results and obtained the highest TOC removal as 95.68% which based on the regression model; and also with target of the highest TOC removal rate, Matlab genetic algorithm toolbox was combined to optimize the TOC removal rate as 99.70% which based on BP neural network model, with the baking temperature was 500.25 oC, baking time was 5h and metal matching was 0.76. The optimization result obtained by genetic algorithm based on BP neural network was 10.58% higher than the average experimental ones, and also slightly higher than the one achieved by regression model.Secondly, multiple non-linear regression mathematical model and BP neural network method were established to investigate the relationships between four main factors （TOC concentration of butyric acid, amount of catalyst, reaction temperature, oxygen partial pressure） and degradation result of butyric acid respectively, according to the experimental data of heating process. The average relative deviation calculated of regression model was slightly lower than the one of BP model. The degradation cost optimized by Lingo was significantly lower than the actual experiment one with the same degradation effect; the highest TOC removal rate was set as the target of optimization, and find out the best solution with Matlab genetic algorithm toolbox. The results optimized by each method had not received a significant difference with the same initial concentration and degradation cost; they had played great effect in saving cost and optimizing degradation result. The cost optimized by Lingo based on regression model was lower than the experimental one with the same degradation effect, and find out the best TOC removal rate through Matlab genetic algorithm toolbox based on BP model. The result optimized by both two methods had not significant difference with the same initial concentration and degradation cost, and both of them played great roles in degradation result and cost.Thirdly, mathematical model with multiple non-linear regression and BP neural network method was established to investigate the relationship between four main factors （TOC concentration of butyric acid, the amount of catalyst, reaction temperature, oxygen partial pressure, reaction time） and degradation result of butyric acid respectively, according to the experimental data of constant temperature process in butyric acid CWAO degradation. The cost optimized by Lingo software based on the regression model was significantly less than the experimental one with the same degradation results. Meanwhile, the degradation effect optimized by genetic algorithm based on BP artificial neural network was significantly better than the one optimized by Lingo with the same degradation cost.Finally, BP neural network was applied to simulate and optimize the meta-nitro benzene sulfonic acid sodium salt Catalytic Wet Hydrogen Peroxide Oxidation （CWPO） degradation process. The model was established to investigate the effect on degradation process which influenced by the amount of catalyst, initial pressure of oxygen, initial concentration of meta-nitro benzene sulfonic acid sodium salt, amount of H2O2, reaction temperature, reaction time and pH, and get the factors’order in significance on the TOC removal rate by factors’sensitivity analysis, the influence degree of these factors can be shown as follow: reaction temperature > initial concentration of meta-nitro benzene sulfonic acid sodium salt > pH > amount of H2O2 > reaction time > initial pressure of oxygen > amount of catalyst. With the highest TOC removal rate as the optimization objective, the degradation conditions optimization with non-cost constraints and cost constraints were carried out, the mean TOC removal rate optimized by cost constrains was more than 10% higher than the average of the experimental ones, and the mean cost was 0.2 yuan less than the one optimized by non-cost constrains.The icfluences of degradation effect with different reaction conditions during the processes of Mn-Ce catalyst preparation and butyric acid CWAO degradation can be described by methods of non-linear regression and artificial neural network, it is help to reduce the wastage, save time and it plays great significance to guide practice by seeking out the optimal experimental conditions with the corresponding optimization methods.更多还原