不同進水方式潮汐流人工濕地污染物去除研究
發(fā)布時間：2019-01-08 17:54
【摘要】：目前,水污染問題日益嚴(yán)峻,自然濕地自身凈化能力有限,人工濕地具有處理效果好、費用低的優(yōu)勢,逐漸成為研究的熱點,但是其復(fù)氧能力較低,并且受多種因素影響,存在一定的缺陷,因此需要尋求強化人工濕地的技術(shù)方法。潮汐流人工濕地(Tidal Flow Constructed Wetland,TF-CW)是一種新型人工濕地生態(tài)系統(tǒng),并且在污染物去除方面受到了廣泛的關(guān)注。本文為研究TF-CW中污染物去除效果及影響因素,建立潮汐流模擬裝置,通過對比不同進水方式下(連續(xù)流濕地(A),設(shè)閑置/反應(yīng)時間分別為1:1(B),1:2(C),2:1(D))的模擬裝置對污染物的去除效果以及沿程變化規(guī)律,并用冗余分析(RDA)篩選出影響其去除效果的主要因子,將各個影響因子輸入人工神經(jīng)網(wǎng)絡(luò)模型進行各個污染指標(biāo)出水濃度的訓(xùn)練與驗證。得出如下結(jié)論:1.A、B、C、D四種進水方式對TN的平均去除率分別為82.41%±4.84%、84.82%±5.09%、86.09%±3.99%、90.23%±3.05%。四種進水方式差異顯著(P0.05);A進水方式NH4+-N去除效果與B、C、D差異顯著(P0.05),其中D進水方式NH4+-N的去除效果最好,但A對NO3--N的總體去除效果較優(yōu);四種進水方式對TP的去除率差異性均不顯著(P0.05);閑置/反應(yīng)時間并不影響TOC(總有機碳)的去除率。四種進水方式下,NH4+-N去除率均在0~15cm深度內(nèi)最大,隨深度增加,去除率下降;NO3--N濃度在0~15cm深度內(nèi)迅速上升;隨處理深度增加,TP濃度逐步降低。TOC濃度處于0~20mg/L間的較低水平,并隨深度增加而下降。2.四種進水方式下TF-CW的平均硝化強度差異顯著(P0.05),其中A與其他三種潮汐進水方式均差異顯著,而D是基質(zhì)平均硝化強度最大的進水方式;四種模擬裝置的基質(zhì)平均反硝化強度差異性也顯著(P0.05),A進水方式反硝化強度最大。TF-CW基質(zhì)硝化強度與NH4+-N的去除率之間存在明顯的正相關(guān)性(R2=0.8497,P0.05);反硝化強度與NO3--N的出水濃度呈呈明顯負相關(guān)關(guān)系(R2=0.8448,P0.05)。裝置上部0~30cm的處理深度硝化強度最大,反硝化強度則在中部的30~60cm階段較高。3.RDA分析結(jié)果顯示TN的去除率的主要影響因子有DO(溶解氧)、RAT(淹沒排空比)、ORP(氧化還原電位)、TOC,NH4+-N的主要影響因子有DO、RAT、ORP、Depth(處理深度),NO3--N的主要影響因子有Cond(電導(dǎo)率)、Temp(水溫)、Sal(鹽度)、p H,TP的主要影響因子有DO、RAT、Time(時間)、Depth。因此在用BP神經(jīng)網(wǎng)絡(luò)對TF-CW水體污染物出水濃度進行模擬時選擇各指標(biāo)的主要影響因子作為輸入層,污染物指標(biāo)出水濃度作為輸出層,經(jīng)試錯法可得TN、NH4+-N、NO3--N和TP選擇的隱含層節(jié)點數(shù)分別為9,11,12,9。對數(shù)據(jù)組進行訓(xùn)練的結(jié)果顯示,BP人工神經(jīng)網(wǎng)絡(luò)模型可以有效地預(yù)測污染物的出水濃度,模型預(yù)測值與實際值存在一定的相關(guān)性,也存在較小范圍的誤差。BP神經(jīng)網(wǎng)絡(luò)對各指標(biāo)的擬合能力TP出水濃度最好,總體R2可達0.90076。對TN、NH4+-N和NO3--N的擬合系數(shù)分別為0.67086、0.72854和0.69293。
[Abstract]:At present, the problem of water pollution is becoming more and more serious, the natural wetland's own purification ability is limited, and the artificial wetland has the advantages of good treatment effect and low cost, so it has gradually become the hot spot of research, but its reoxygenation ability is low, and it is influenced by many factors. There are some defects, so it is necessary to seek the technical method to strengthen the constructed wetland. Tidal flow constructed wetland (Tidal Flow Constructed Wetland,TF-CW) is a new constructed wetland ecosystem. In order to study the removal efficiency and influencing factors of pollutants in TF-CW, a tidal flow simulation device was established, and the idle / reaction time of 1:1 (B), was set up by comparing different influent modes (A), of continuous flow wetland). At 1:2 (C), 2:1 (D) simulator, the pollutant removal efficiency and its variation along the path were obtained. The main factors affecting the removal efficiency were screened by redundancy analysis (RDA). The influence factors were inputted into the artificial neural network model to train and verify the effluent concentration of each pollution index. The results are as follows: 1. The average removal rate of TN in the four influent modes is 82.41% 鹵4.84% 鹵84.82% 鹵5.092.92% 鹵3.990.23% 鹵3.05, respectively. There were significant differences among the four influent modes (P0.05). The removal efficiency of NH4 N in A influent mode was significantly different from that in NH4 D (P0.05). The removal efficiency of NH4 N in D influent mode was the best, but the overall removal efficiency of NO3--N was better in A; There was no significant difference in the removal efficiency of TP among the four influent methods (P0.05), while idle / reaction time had no effect on the removal rate of TOC (total organic carbon). Under four influent conditions, the removal rate of NH4-N was the highest in the depth of 0~15cm, and the removal rate decreased with the increase of the depth, and the concentration of NO3--N increased rapidly in the depth of 0~15cm. With the increase of treatment depth, the concentration of TP gradually decreased, while the concentration of TOC was lower than that of 0~20mg/L, and decreased with the increase of depth. The average nitrification intensity of TF-CW under four influent modes was significantly different (P0.05), among which A was significantly different from the other three tidal influent modes, while D was the influent mode with the highest average nitrification intensity of substrate. The difference of the average denitrification intensity of the four simulators was also significant (P0.05) the denitrification intensity of), A was the largest. There was a significant positive correlation between the nitrification intensity of TF-CW substrate and the removal rate of NH4-N (R2 + 0.8497). P0.05); There was a negative correlation between denitrification intensity and effluent concentration of NO3--N (P 0.05). The treatment depth nitrification intensity of 0~30cm in the upper part of the unit was the highest, while the denitrification intensity was higher in the 30~60cm stage in the middle part of the unit. The results of 3.RDA analysis showed that the main influencing factor of TN removal efficiency was DO (dissolved oxygen), RAT (inundated emptying ratio). ORP (redox potential), DO,RAT,ORP,Depth (treatment depth) and Cond (), Temp (water temperature,), Sal (salinity), p H,) are the main influencing factors of TOC,NH4 N and NO3--N. The main influencing factor of TP is DO,RAT,Time (time), Depth.) Therefore, when BP neural network is used to simulate the effluent concentration of TF-CW pollutants, the main influencing factors of each index are selected as the input layer, the effluent concentration of the pollutants is taken as the output layer, and the TN,NH4 N can be obtained by trial and error method. The number of hidden layer nodes selected by NO3--N and TP is 9 / 11 / 12 / 9 respectively. The results of training the data group show that the BP artificial neural network model can effectively predict the effluent concentration of pollutants, and the predicted value of the model has a certain correlation with the actual value. The BP neural network has the best fit ability for each index, and the overall R2 is 0.90076. The fitting coefficients for TN,NH4 N and NO3--N were 0.67086, 0.72854 and 0.69293, respectively.
【學(xué)位授予單位】：中國林業(yè)科學(xué)研究院
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：X703
[Abstract]:At present, the problem of water pollution is becoming more and more serious, the natural wetland's own purification ability is limited, and the artificial wetland has the advantages of good treatment effect and low cost, so it has gradually become the hot spot of research, but its reoxygenation ability is low, and it is influenced by many factors. There are some defects, so it is necessary to seek the technical method to strengthen the constructed wetland. Tidal flow constructed wetland (Tidal Flow Constructed Wetland,TF-CW) is a new constructed wetland ecosystem. In order to study the removal efficiency and influencing factors of pollutants in TF-CW, a tidal flow simulation device was established, and the idle / reaction time of 1:1 (B), was set up by comparing different influent modes (A), of continuous flow wetland). At 1:2 (C), 2:1 (D) simulator, the pollutant removal efficiency and its variation along the path were obtained. The main factors affecting the removal efficiency were screened by redundancy analysis (RDA). The influence factors were inputted into the artificial neural network model to train and verify the effluent concentration of each pollution index. The results are as follows: 1. The average removal rate of TN in the four influent modes is 82.41% 鹵4.84% 鹵84.82% 鹵5.092.92% 鹵3.990.23% 鹵3.05, respectively. There were significant differences among the four influent modes (P0.05). The removal efficiency of NH4 N in A influent mode was significantly different from that in NH4 D (P0.05). The removal efficiency of NH4 N in D influent mode was the best, but the overall removal efficiency of NO3--N was better in A; There was no significant difference in the removal efficiency of TP among the four influent methods (P0.05), while idle / reaction time had no effect on the removal rate of TOC (total organic carbon). Under four influent conditions, the removal rate of NH4-N was the highest in the depth of 0~15cm, and the removal rate decreased with the increase of the depth, and the concentration of NO3--N increased rapidly in the depth of 0~15cm. With the increase of treatment depth, the concentration of TP gradually decreased, while the concentration of TOC was lower than that of 0~20mg/L, and decreased with the increase of depth. The average nitrification intensity of TF-CW under four influent modes was significantly different (P0.05), among which A was significantly different from the other three tidal influent modes, while D was the influent mode with the highest average nitrification intensity of substrate. The difference of the average denitrification intensity of the four simulators was also significant (P0.05) the denitrification intensity of), A was the largest. There was a significant positive correlation between the nitrification intensity of TF-CW substrate and the removal rate of NH4-N (R2 + 0.8497). P0.05); There was a negative correlation between denitrification intensity and effluent concentration of NO3--N (P 0.05). The treatment depth nitrification intensity of 0~30cm in the upper part of the unit was the highest, while the denitrification intensity was higher in the 30~60cm stage in the middle part of the unit. The results of 3.RDA analysis showed that the main influencing factor of TN removal efficiency was DO (dissolved oxygen), RAT (inundated emptying ratio). ORP (redox potential), DO,RAT,ORP,Depth (treatment depth) and Cond (), Temp (water temperature,), Sal (salinity), p H,) are the main influencing factors of TOC,NH4 N and NO3--N. The main influencing factor of TP is DO,RAT,Time (time), Depth.) Therefore, when BP neural network is used to simulate the effluent concentration of TF-CW pollutants, the main influencing factors of each index are selected as the input layer, the effluent concentration of the pollutants is taken as the output layer, and the TN,NH4 N can be obtained by trial and error method. The number of hidden layer nodes selected by NO3--N and TP is 9 / 11 / 12 / 9 respectively. The results of training the data group show that the BP artificial neural network model can effectively predict the effluent concentration of pollutants, and the predicted value of the model has a certain correlation with the actual value. The BP neural network has the best fit ability for each index, and the overall R2 is 0.90076. The fitting coefficients for TN,NH4 N and NO3--N were 0.67086, 0.72854 and 0.69293, respectively.
【學(xué)位授予單位】：中國林業(yè)科學(xué)研究院
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：X703
【相似文獻】
相關(guān)期刊論文 前7條
1 關(guān)宗山,安洪儒;棒(線)冷卻裝置的水力學(xué)研究[J];冶金設(shè)備;1981年05期
2 周健;竇艷艷;何強;栗靜靜;陳W，
本文編號：2404937
本文鏈接：http://sikaile.net/kejilunwen/huanjinggongchenglunwen/2404937.html
最近更新
教材專著
